On 05/05/11 22:11, Mariano Martinez Peck wrote:

http://stephane.ducasse.free.fr/FreeBooks.html

And of those many good books I'd put in a particular word for "Smalltalk By Example" by Alex Sharp. I found it particularly straightforward and helpful when I was starting out. http://stephane.ducasse.free.fr/FreeBooks/ByExample/ Steve

On 05/05/2011 05:26 PM, Cédrick Béler wrote:

...

Lastly, most Smalltalk systems are image based...

...which makes you feel the system is "alive", hence one **huge benefit** of Smalltalk: its debugger which enables on the fly debbuging... and also test driven development (real one [1]) where you can run incomplete code and code what's missing iteratively when you need it (Smalltalk is a live system, not only a language as somebody said lately).

Cédrick

[1] see in particular this webcast: http://www.pharocasts.com/2010/01/starting-with-sunit-and-debugger.html

Seriously ... these points in favor of the image are so m00t. Lets see how it would work without an image: I write a C application which I link to GCC. Now I run GDB on my application, and while running I have the whole GCC compiler collection at my disposal while running. While debugging (at some breakpoint) I just let the GCC library compile some C code for me; I turn on the executable flag and whooptidoo, I have a Smalltalk like debugger for C. This is totally unrelated to having an image; it's just a great debugger implementation. 2 completely different things. No reason why this wouldn't work for C; except for the fact that they didn't do it yet (those lazy bastards).

On 2011-05-05, at 17:39, Cédrick Béler wrote:

Le 5 mai 2011 à 17:32, Toon Verwaest a écrit :

...

On 05/05/2011 05:26 PM, Cédrick Béler wrote:

...

...

Lastly, most Smalltalk systems are image based...

...which makes you feel the system is "alive", hence one **huge benefit** of Smalltalk: its debugger which enables on the fly debbuging... and also test driven development (real one [1]) where you can run incomplete code and code what's missing iteratively when you need it (Smalltalk is a live system, not only a language as somebody said lately).

Cédrick

[1] see in particular this webcast: http://www.pharocasts.com/2010/01/starting-with-sunit-and-debugger.html

Seriously ... these points in favor of the image are so m00t. Lets see how it would work without an image:

I write a C application which I link to GCC. Now I run GDB on my application, and while running I have the whole GCC compiler collection at my disposal while running. While debugging (at some breakpoint) I just let the GCC library compile some C code for me; I turn on the executable flag and whooptidoo, I have a Smalltalk like debugger for C.

This is totally unrelated to having an image; it's just a great debugger implementation. 2 completely different things. No reason why this wouldn't work for C; except for the fact that they didn't do it yet (those lazy bastards).

ok, true :)

but, it's not only the debugger... getting senders, implementers, class that use it, methods that contains this word, ... of course, this is doable with files (see eclipse *sigh*), but I prefer the snappy feeling of an image for that...

the snappiness could be easily achieved with a fulltext search index (just look how fast spotlight works on a couple of millions of files on my machine...) The main issue of the image, is that it locks you out, making it close to impossible to use other tools than smalltalk. Though the nice thing about the image is that you can start up your IDE in less than a second, and again, this is not necessarily linked to Smalltalk and could be achieved for C programs as well.

On 05/05/2011 05:46 PM, Dave Mason wrote:

On May 5, 2011, at 11:32, Toon Verwaest wrote:

...

This is totally unrelated to having an image; it's just a great debugger implementation. 2 completely different things. No reason why this wouldn't work for C; except for the fact that they didn't do it yet (those lazy bastards). ... and that it's an order of magnitude more difficult!

Yes, there are certainly other aspects of being image-based that are way cool, but iterative code development and debugging is a big part of it.

(I am teaching a Java course right now with BlueJ, which is a cool environment (at least if you're not used to an image-based system like Smalltalk). It provides a little UML diagram of your classes, and you can create an object by clicking on the appropriate class, and inspect, and call methods. Objects that you create appear on a "workbench". So I had build up a little example with a few objects, found a bug, and edited the class, hit "compile" and *poof* empty workbench. Yes, in Java they probably could have emulated the image, but it's a lot of work!)

../Dave It's more difficult because it's not integrated yet. Suppose that this was already integrated within GDB? Suppose GDB looked like a Smalltalk debugger? Is there anything technical that stands in your way?

C is annoying for many other reasons, but not for that one.

On 5 May 2011 18:57, Jimmie Houchin <jlhouchin@gmail.com> wrote:

On 5/5/2011 10:32 AM, Toon Verwaest wrote:

...

On 05/05/2011 05:26 PM, Cédrick Béler wrote:

...

...

Lastly, most Smalltalk systems are image based...

...which makes you feel the system is "alive", hence one **huge benefit** of Smalltalk: its debugger which enables on the fly debbuging... and also test driven development (real one [1]) where you can run incomplete code and code what's missing iteratively when you need it (Smalltalk is a live system, not only a language as somebody said lately).

Cédrick

[1] see in particular this webcast: http://www.pharocasts.com/2010/01/starting-with-sunit-and-debugger.html

Seriously ... these points in favor of the image are so m00t. Lets see how it would work without an image:

I write a C application which I link to GCC. Now I run GDB on my application, and while running I have the whole GCC compiler collection at my disposal while running. While debugging (at some breakpoint) I just let the GCC library compile some C code for me; I turn on the executable flag and whooptidoo, I have a Smalltalk like debugger for C.

This is totally unrelated to having an image; it's just a great debugger implementation. 2 completely different things. No reason why this wouldn't work for C; except for the fact that they didn't do it yet (those lazy bastards).

Disclaimer, I am not a professional programmer and I have not programmed in C or any other static compiled language at all. So I may not have any idea what I am talking about. :)

What I don't see in your example, but what I experience in my programming is often in the *use* of my application during development. Part of my application is written in Python (so I can interface I library I can't from Pharo) and part in Pharo.  (*use*, ie:  the application is compiled and ready for a *user* and I am the user and encounter the bug.

My Python app connects to a server on the net and downloads data and inserts it into a PostgreSQL database. It also provides live data to my Pharo app via an http interface.

While experimenting with NumPy and running some statistical analysis on a GB or two of data. The analysis took 10-12 hours to run. I start it up and walk away. I get back to it only to find that at the end of all the analysis it exited with a stacktrace. Ugh!  I concatenated a string using a ',' instead of a '+'.   Switching between Pharo and Python bit me. :)  This was at the very end of all the analysis and the generation of a report. The last line of the code had the bug. I lost the 12hours of analysis. This bug passed the syntax checker and the Python compiler when I imported the module. It wasn't until it was live and running and encountered in execution that it was discovered.

In Pharo, the same situation, I fix it in the debugger and my report is generated and I continue. All my data is still resident in the image and I can explore and continue. In Python (and I presume most any other dead language) all is lost and I have to rerun.

Now as I said I am not a professional. Certain processes would have helped in this situation, but not necessarily all. And I have limited experience with the best other systems or languages have to offer.

But how many programming languages/environments handle that well?

I am still so very tired of every time an application on my computer says, "You just updated. Do you want to restart?"

Why do most every single web application outside of the Smalltalk world require shutdowns and restarts for bug fixes and upgrades?

Why does NetBeans consume 200mb, take 3-5minutes to startup, and do a much, much poorer job with syntax, autocomplete, etc. on a single 500 line Python file?

I don't think the world out there is anywhere near what Smalltalk has. Nor do they have the vision for it. They are happy with files, and restarts and all their processes to reduce their pain due to not having a live environment.

I am a single guy with a vision, an idea, that I want to implement for my business. I have seen nothing that gives me the productivity that Smalltalk does. Not Java, Python, Clojure, Scala, Lua, etc.

I have never seen any other language accomplish so much with so few. I look at other languages and apps built with them and look at the armies they have to do so. Then I look at Seaside, etc. with just a small group of guys with an idea of something better.

While I desperately want Squeak/Pharo to be able to interface outside systems better. I have seen nothing that is better or comes close. And as I said, I have a business requirement to interface a proprietary library written in C, which why I am connected with Python (and Windows) against my will. So I do understand.

I understand, as does everybody else here, that Pharo/Smalltalk is not perfect and has huge areas where we have not done things well or right. But I don't think this is one of them. I think this is one where the rest of the world is behind us.

Squeak/Pharo/Smalltalk is empowering.

Well said. Except that i'm not sharing your view that its hard to interface with foreign libraries. Its not hard at all. Of course to connect two different worlds, you need to have knowledge in both of them. But this requirement not a bit different when you take any other pair of languages and try to connect them.

JMHO, take with a grain of salt. :)

Jimmie

-- Best regards, Igor Stasenko AKA sig.

Thanks Jimmie Thanks to kick our asses like that! I would love to have that too in Pharo. Stef

Thanks,

I have no knowledge of either the knowledge or the challenges involved in using external libraries in Pharo or Squeak. I have no knowledge of FFI/Alien or using C/C++/C# or compilers.

However, this is my experience in Python.

Navigate to the directory containing the script makepy.py or if it is a part of your Python's sys.path, execute the script. It generates a Python module which is on

It pops up a dialogue which prompts you to select the library you wish to interface.

Then to use in a script simply import Dispatch self.mylib = Dispatch("MyLibrary")

This will expose all the functionality of the library.

All provided by the python win32 extensions. It was very successful for my needs. I do not know what limitations it may or may not have.

Very easy for non-expert programmers. I would love this level of ability to interface outside libraries in Squeak. But I have no idea the effort required to automate the generation of a class or classes which interface the external library.

In my particular instance this is obviously for a Windows library. I don't know if Python has anything comparable for Linux or OSX.

In this particular instance, Python was enabling for me, for which I am grateful. Otherwise I might be stuck writing my app in VisualBasic. But despite my gratefulness, I spend as little time in Python as possible. Despite Python not requiring a compiler, I really hate going to an editor and writing code. Then to an interpreter to run code. Hit my stacktrace. Go back to the editor. Reload the module in the interpreter and run again, and if that doesn't succeed due to the reload not really reloading the new code, open in a new interpreter. Ugh!!! Where's my Smalltalk. Give my live object system. :)

Jimmie

On 5/5/2011 4:25 PM, Igor Stasenko wrote:

On 5 May 2011 21:39, Jimmie Houchin<jlhouchin@gmail.com> wrote:

...

On 5/5/2011 12:21 PM, Igor Stasenko wrote:

...

On 5 May 2011 18:57, Jimmie Houchin<jlhouchin@gmail.com> wrote: [Big Snip] Well said. Except that i'm not sharing your view that its hard to interface with foreign libraries. Its not hard at all. Of course to connect two different worlds, you need to have knowledge in both of them. But this requirement not a bit different when you take any other pair of languages and try to connect them. Thanks,

I have no knowledge of either the knowledge or the challenges involved in using external libraries in Pharo or Squeak. I have no knowledge of FFI/Alien or using C/C++/C# or compilers.

However, this is my experience in Python.

Navigate to the directory containing the script makepy.py or if it is a part of your Python's sys.path, execute the script. It generates a Python module which is on

It pops up a dialogue which prompts you to select the library you wish to interface.

Then to use in a script simply import Dispatch self.mylib = Dispatch("MyLibrary")

This will expose all the functionality of the library.

All provided by the python win32 extensions. It was very successful for my needs. I do not know what limitations it may or may not have.

Very easy for non-expert programmers. I would love this level of ability to interface outside libraries in Squeak. But I have no idea the effort required to automate the generation of a class or classes which interface the external library.

In my particular instance this is obviously for a Windows library. I don't know if Python has anything comparable for Linux or OSX.

In this particular instance, Python was enabling for me, for which I am grateful. Otherwise I might be stuck writing my app in VisualBasic. But despite my gratefulness, I spend as little time in Python as possible. Despite Python not requiring a compiler, I really hate going to an editor and writing code. Then to an interpreter to run code. Hit my stacktrace. Go back to the editor. Reload the module in the interpreter and run again, and if that doesn't succeed due to the reload not really reloading the new code, open in a new interpreter. Ugh!!! Where's my Smalltalk. Give my live object system. :) Haha.. i have strong suspicion that here you are talking about quite specific set of libraries, which using OLE/COM interfaces. Indeed, one could implement an automatic "import/connect" tool for it, because a library itself contain enough information reflecting it interface(s). You can check Dolphin smalltalk which works only on windows and has integrated solution for that for years: In same way like you described, you just pick the library, click "ok" and its done& ready for use.

But that would be too easy if you could do same with any other library in your system. So, if you feel adventurous, try to check what Python can do with libraries like Kernel32.dll or User32.dll. I am sure it can do as little as Smalltalk in this regard, but you are free to check it yourself :)

As I said, I am totally unaware of the limitations of the system, but that it did what I need, and would have liked to do that from Pharo/Squeak. I know Dolphin has/had certain capabilities. But I don't prefer to use non-open source software if at all possible for development. I also am very preferential towards cross-platform software. Dolphin fails on all accounts. I would choose my Python/Pharo blend over Dolphin any day. I know it introduces some pain, but I am willing to accept the pain. I like tools that allow me to use them where ever I am and whatever I am doing. However, that said, when I look at the facilities Python offers for such capabilities it on appearances looks quite impressive. http://docs.python.org/library/ctypes.html """ ctypes is a foreign function library for Python. It provides C compatible data types, and allows calling functions in DLLs or shared libraries. It can be used to wrap these libraries in pure Python. 15.18.1.1. Loading dynamic link libraries ctypes exports the /cdll/, and on Windows /windll/ and /oledll/ objects, for loading dynamic link libraries. You load libraries by accessing them as attributes of these objects. /cdll/ loads libraries which export functions using the standard cdecl calling convention, while /windll/ libraries call functions using the stdcall calling convention. /oledll/ also uses the stdcall calling convention, and assumes the functions return a Windows HRESULT error code. The error code is used to automatically raise a WindowsError exception when the function call fails. Here are some examples for Windows. Note that msvcrt is the MS standard C library containing most standard C functions, and uses the cdecl calling convention:

...

...

from ctypes import * print windll.kernel32 # doctest: +WINDOWS <WinDLL 'kernel32', handle ... at ...> print cdll.msvcrt # doctest: +WINDOWS <CDLL 'msvcrt', handle ... at ...> libc = cdll.msvcrt # doctest: +WINDOWS

""" It has examples for windows and linux. I didn't read it all as I am unqualified to assess its capabilities or limitations, nor do I presently need it as the win32 extensions do easily and well what I presently need to do. Hopefully it can inform us as to what is expected and doable in alternative languages which are dynamic like Smalltalk. Jimmie

On Fri, May 6, 2011 at 3:19 AM, Igor Stasenko <siguctua@gmail.com> wrote:

On 6 May 2011 01:03, Jimmie Houchin <jlhouchin@gmail.com> wrote:

...

On 5/5/2011 4:25 PM, Igor Stasenko wrote:

On 5 May 2011 21:39, Jimmie Houchin <jlhouchin@gmail.com> wrote:

On 5/5/2011 12:21 PM, Igor Stasenko wrote:

On 5 May 2011 18:57, Jimmie Houchin<jlhouchin@gmail.com> wrote:

[Big Snip]

Well said. Except that i'm not sharing your view that its hard to interface with foreign libraries. Its not hard at all. Of course to connect two different worlds, you need to have knowledge in both of them. But this requirement not a bit different when you take any other pair of languages and try to connect them.

Thanks,

I have no knowledge of either the knowledge or the challenges involved in using external libraries in Pharo or Squeak. I have no knowledge of FFI/Alien or using C/C++/C# or compilers.

So, read about it. It doesn't bites. :)

...

As I said, I am totally unaware of the limitations of the system, but that it did what I need, and would have liked to do that from Pharo/Squeak. I know Dolphin has/had certain capabilities. But I don't prefer to use non-open source software if at all possible for development. I also am very preferential towards cross-platform software. Dolphin fails on all accounts. I would choose my Python/Pharo blend over Dolphin any day. I know it introduces some pain, but I am willing to accept the pain. I like tools that allow me to use them where ever I am and whatever I am doing.

However, that said, when I look at the facilities Python offers for such capabilities it on appearances looks quite impressive.

http://docs.python.org/library/ctypes.html

""" ctypes is a foreign function library for Python. It provides C compatible data types, and allows calling functions in DLLs or shared libraries. It can be used to wrap these libraries in pure Python.

15.18.1.1. Loading dynamic link libraries

ctypes exports the cdll, and on Windows windll and oledll objects, for loading dynamic link libraries.

You load libraries by accessing them as attributes of these objects. cdll loads libraries which export functions using the standard cdecl calling convention, while windll libraries call functions using the stdcall calling convention. oledll also uses the stdcall calling convention, and assumes the functions return a Windows HRESULT error code. The error code is used to automatically raise a WindowsError exception when the function call fails.

Here are some examples for Windows. Note that msvcrt is the MS standard C library containing most standard C functions, and uses the cdecl calling convention:

...

...

...

from ctypes import * print windll.kernel32 # doctest: +WINDOWS <WinDLL 'kernel32', handle ... at ...> print cdll.msvcrt # doctest: +WINDOWS <CDLL 'msvcrt', handle ... at ...> libc = cdll.msvcrt # doctest: +WINDOWS

So? Why do you think that you cannot do the same in Pharo? Have you digested what is available before doing it in Python?

...

"""

It has examples for windows and linux.

I feel that you are seriously under-informed. Just try googling for:

Squeak + FFI Alien

NativeBoost

and just check what you can do using it:

http://www.squeaksource.com/NBOpenGL/

also check mail archives.

Forgive me, but i really can't understand , why you can't just look for what you need by yourself? Are google.com banned by your ISP?

For me this subject is interested in aspect, why this information (while being available openly) didn't catch your attention , so you using python instead. (Okay, you might miss some functionality like being able to automatically generate bindings to COM interfaces). But again, i am sure that if you surf the net or ask right questions on mailing list, you will probably discover that there is already work being done in that direction.

Indeed, one thing FFI / Alien / NB don't have is a documentation as nice as http://docs.python.org/library/ctypes.html in http://book.pharo-project.org/ FFI / Alien / NB may be powerful, it seems I've never managed to do what I want, even with all mailing lists support (search mails on yaz / z3950 or id3taglib). In Python it works on OSX out of the box (I've just tried :) But when I was working with Python and Ruby several years ago things was easy (for my needs), I think because of documentation. Laurent And at last, if you feel that there is missing some key functionality

which you want, and you expect that it is also could be useful to community, you can always implement it and share with people (instead of using Python ;). So, next who will come after you won't find himself in a desert with couple of oasises few hundred miles away.

...

I didn't read it all as I am unqualified to assess its capabilities or limitations, nor do I presently need it as the win32 extensions do easily and well what I presently need to do.

Hopefully it can inform us as to what is expected and doable in alternative languages which are dynamic like Smalltalk.

Jimmie

-- Best regards, Igor Stasenko AKA sig.

On Fri, May 6, 2011 at 4:01 PM, Igor Stasenko <siguctua@gmail.com> wrote:

On 6 May 2011 07:11, laurent laffont <laurent.laffont@gmail.com> wrote:

...

On Fri, May 6, 2011 at 3:19 AM, Igor Stasenko <siguctua@gmail.com>

wrote:

...

...

On 6 May 2011 01:03, Jimmie Houchin <jlhouchin@gmail.com> wrote:

...

On 5/5/2011 4:25 PM, Igor Stasenko wrote:

On 5 May 2011 21:39, Jimmie Houchin <jlhouchin@gmail.com> wrote:

On 5/5/2011 12:21 PM, Igor Stasenko wrote:

On 5 May 2011 18:57, Jimmie Houchin<jlhouchin@gmail.com> wrote:

[Big Snip]

Well said. Except that i'm not sharing your view that its hard to interface with foreign libraries. Its not hard at all. Of course to connect two different worlds, you need to have knowledge in both of them. But this requirement not a bit different when you take any other pair of languages and try to connect them.

Thanks,

I have no knowledge of either the knowledge or the challenges involved in using external libraries in Pharo or Squeak. I have no knowledge of FFI/Alien or using C/C++/C# or compilers.

So, read about it. It doesn't bites. :)

...

As I said, I am totally unaware of the limitations of the system, but that it did what I need, and would have liked to do that from Pharo/Squeak.

I

...

...

know Dolphin has/had certain capabilities. But I don't prefer to use non-open source software if at all possible for development. I also am very preferential towards cross-platform software. Dolphin fails on all accounts. I would choose my Python/Pharo blend over Dolphin any day. I know it introduces some pain, but I am willing to accept the pain. I like tools that allow me to use them where ever I am and whatever I am doing.

However, that said, when I look at the facilities Python offers for such capabilities it on appearances looks quite impressive.

http://docs.python.org/library/ctypes.html

""" ctypes is a foreign function library for Python. It provides C compatible data types, and allows calling functions in DLLs or shared libraries. It can be used to wrap these libraries in pure Python.

15.18.1.1. Loading dynamic link libraries

ctypes exports the cdll, and on Windows windll and oledll objects, for loading dynamic link libraries.

You load libraries by accessing them as attributes of these objects. cdll loads libraries which export functions using the standard cdecl calling convention, while windll libraries call functions using the stdcall calling convention. oledll also uses the stdcall calling convention, and assumes the functions return a Windows HRESULT error code. The error code is used to automatically raise a WindowsError exception when the function call fails.

Here are some examples for Windows. Note that msvcrt is the MS standard C library containing most standard C functions, and uses the cdecl calling convention:

...

...

...

from ctypes import * print windll.kernel32 # doctest: +WINDOWS <WinDLL 'kernel32', handle ... at ...> print cdll.msvcrt # doctest: +WINDOWS <CDLL 'msvcrt', handle ... at ...> libc = cdll.msvcrt # doctest: +WINDOWS

So? Why do you think that you cannot do the same in Pharo? Have you digested what is available before doing it in Python?

...

"""

It has examples for windows and linux.

I feel that you are seriously under-informed. Just try googling for:

Squeak + FFI Alien

NativeBoost

and just check what you can do using it:

http://www.squeaksource.com/NBOpenGL/

also check mail archives.

Forgive me, but i really can't understand , why you can't just look for what you need by yourself? Are google.com banned by your ISP?

For me this subject is interested in aspect, why this information (while being available openly) didn't catch your attention , so you using python instead. (Okay, you might miss some functionality like being able to automatically generate bindings to COM interfaces). But again, i am sure that if you surf the net or ask right questions on mailing list, you will probably discover that there is already work being done in that direction.

Indeed, one thing FFI / Alien / NB don't have is a documentation as nice as http://docs.python.org/library/ctypes.html in http://book.pharo-project.org/ FFI / Alien / NB may be powerful, it seems I've never managed to do what I want, even with all mailing lists support (search mails on yaz / z3950 or id3taglib). In Python it works on OSX out of the box (I've just tried :)

Is it works because someone already provided a bindings for libraries you mention, or you did it by yourself?

Have just used ctypes:

...

...

import ctypes import ctypes.util ctypes.util.find_library("yaz") '/usr/local/lib/libyaz.dylib' yaz = ctypes.cdll.LoadLibrary("libyaz.dylib") con = yaz.ZOOM_connection_new("z3950.loc.gov", 7090);

Laurent

Here a first hit from google search:

http://lists.indexdata.dk/pipermail/yazlist/2002-May/000269.html

<quote> I wrote these files for more than 2 years ago to play with SWIG and PERL, they are far from being used, but I have just done a fast perl test and the results were just fine.

zclient.c zclient.h zclinet.i swig interface file

The zclient_wrap.c will be created as

...

swig -python zclient.i

which should be compiled and linked with the rest </quote>

So? Do you think it was easy in first place to a guy who did it from scratch? Now you can easily use it (and its just works out of the box). But it is because someone did it before you!

And if you gonna repeat same from scratch, then you will have to learn: - C - SWIG - Python

so, do you still think that it easier to do that in Python than in Smalltalk?

...

But when I was working with Python and Ruby several years ago things was easy (for my needs), I think because of documentation. Laurent

-- Best regards, Igor Stasenko AKA sig.

Laurent igor is right if the bindings are done and package then of course this is different than just calling them. Stef

Is it works because someone already provided a bindings for libraries you mention, or you did it by yourself?

Have just used ctypes:

...

...

...

import ctypes import ctypes.util ctypes.util.find_library("yaz") '/usr/local/lib/libyaz.dylib' yaz = ctypes.cdll.LoadLibrary("libyaz.dylib") con = yaz.ZOOM_connection_new("z3950.loc.gov", 7090);

Indeed, I think the great thing is that I could write this in 5 mn just looking at the doc, haven't used python for years.

In Pharo, well, .........

Laurent.

Laurent

Here a first hit from google search:

http://lists.indexdata.dk/pipermail/yazlist/2002-May/000269.html

<quote> I wrote these files for more than 2 years ago to play with SWIG and PERL, they are far from being used, but I have just done a fast perl test and the results were just fine.

zclient.c zclient.h zclinet.i swig interface file

The zclient_wrap.c will be created as

...

swig -python zclient.i

which should be compiled and linked with the rest </quote>

So? Do you think it was easy in first place to a guy who did it from scratch? Now you can easily use it (and its just works out of the box). But it is because someone did it before you!

And if you gonna repeat same from scratch, then you will have to learn: - C - SWIG - Python

so, do you still think that it easier to do that in Python than in Smalltalk?

...

But when I was working with Python and Ruby several years ago things was easy (for my needs), I think because of documentation. Laurent

-- Best regards, Igor Stasenko AKA sig.

On May 6, 2011, at 13:25, Igor Stasenko wrote:

Seriously, if you want such kind of shitty interface, which allows you to call arbitrary function with arbitrary arguments without any argument type checking, i can write it for you.

Well, yes, actually. If it means they can access the big wide world beyond Smalltalk, lots of people *would* like that. Of course they'd like it better if there were the nice type-checking, etc. but presumably that would come incrementally. Seriously, this might be something I could contribute to Pharo. Especially if Igor can get a simple, but functional, start going easily. ../Dave

On May 6, 2011, at 1:25 PM, Igor Stasenko wrote:

Seriously, if you want such kind of shitty interface, which allows you to call arbitrary function with arbitrary arguments without any argument type checking, i can write it for you.

call me crazy but: yes i would like that interface. ctypes works very well. i have used it plenty. it gives you the ability to get access to libraries trivially. you dont have to spend a week writing a wrapper for something you might only need for a short period of time. if you need to use that library extensively then you need to spend the time writing a proper library for it. you can call me crazy but i find functionality like this *extraordinarily* useful. both python and ruby have very nice implementations, its kind of a shame we dont yet. were all adults here, i understand that if i do something stupid i will get bad results, but i like having the ability to make that decision. jb -- A man can live and be healthy without killing animals for food; therefore, if he eats meat, he participates in taking animal life merely for the sake of his appetite. And to act so is immoral. -Leo Tolstoy

On Fri, May 6, 2011 at 7:25 PM, Igor Stasenko <siguctua@gmail.com> wrote:

On 6 May 2011 18:45, laurent laffont <laurent.laffont@gmail.com> wrote:

...

On Fri, May 6, 2011 at 4:01 PM, Igor Stasenko <siguctua@gmail.com> wrote:

...

Is it works because someone already provided a bindings for libraries you mention, or you did it by yourself?

Have just used ctypes:

...

...

...

import ctypes import ctypes.util ctypes.util.find_library("yaz") '/usr/local/lib/libyaz.dylib' yaz = ctypes.cdll.LoadLibrary("libyaz.dylib") con = yaz.ZOOM_connection_new("z3950.loc.gov", 7090);

oh cool.. give me a 5 minutes to load external library, look up the random symbol exported by it and make a call to it with random number of arguments, and expect it won't crash.

Works outta box! Cool.

Now serious. How about spending time writing a bindings for that library? Where you have properly organized classes/interfaces, error handling and other stuff. No takers?

Seriously, if you want such kind of shitty interface, which allows you to call arbitrary function with arbitrary arguments without any argument type checking, i can write it for you.

Do it, please do it. Seriously. Thanks Laurent.

...

Laurent

-- Best regards, Igor Stasenko AKA sig.

On Fri, May 6, 2011 at 8:06 PM, Schwab,Wilhelm K <bschwab@anest.ufl.edu>wrote:

No real agenda either way here, except to say that we almost have that in FFI. If you know enough to make the call, do you not know enough to write the FFI wrapper? Just asking. Having stirred the broth, I'll shut up and learn :)

My last attempts: http://forum.world.st/Re-Z3950-on-OSX-module-yaz3-not-found-tp3161332p316133... http://forum.world.st/Re-FFI-crashes-VM-on-Linux-was-Z3950-on-OSX-module-yaz... Any help appreciated. Laurent.

________________________________________ From: pharo-project-bounces@lists.gforge.inria.fr [ pharo-project-bounces@lists.gforge.inria.fr] On Behalf Of laurent laffont [laurent.laffont@gmail.com] Sent: Friday, May 06, 2011 1:47 PM To: Pharo-project@lists.gforge.inria.fr Subject: Re: [Pharo-project] Popularity of Smalltalk in Software Industry

On Fri, May 6, 2011 at 7:25 PM, Igor Stasenko <siguctua@gmail.com<mailto: siguctua@gmail.com>> wrote: On 6 May 2011 18:45, laurent laffont <laurent.laffont@gmail.com<mailto: laurent.laffont@gmail.com>> wrote:

...

On Fri, May 6, 2011 at 4:01 PM, Igor Stasenko <siguctua@gmail.com <mailto:siguctua@gmail.com>> wrote:

...

Is it works because someone already provided a bindings for libraries you mention, or you did it by yourself?

Have just used ctypes:

...

...

...

import ctypes import ctypes.util ctypes.util.find_library("yaz") '/usr/local/lib/libyaz.dylib' yaz = ctypes.cdll.LoadLibrary("libyaz.dylib") con = yaz.ZOOM_connection_new("z3950.loc.gov<http://z3950.loc.gov>", 7090);

oh cool.. give me a 5 minutes to load external library, look up the random symbol exported by it and make a call to it with random number of arguments, and expect it won't crash.

Works outta box! Cool.

Now serious. How about spending time writing a bindings for that library? Where you have properly organized classes/interfaces, error handling and other stuff. No takers?

Seriously, if you want such kind of shitty interface, which allows you to call arbitrary function with arbitrary arguments without any argument type checking, i can write it for you.

Do it, please do it. Seriously. Thanks

Laurent.

...

Laurent

-- Best regards, Igor Stasenko AKA sig.

On Fri, May 6, 2011 at 9:16 PM, Schwab,Wilhelm K <bschwab@anest.ufl.edu>wrote:

Laurent,

I am starting to understand your pain, or at least how it manifests itself in my world. First, the search path for libraries (Unix VM) can be a problem, not because of how it works, but because it does not tell me what it is trying to do. IMHO, any time the vm translates something (library name to search path, COM port name to number or vice versa), it should describe what it did, perhaps only if the attempt to open/load/whatever fails (if a load works, I'm probably happy and not looking). I have been known to hack the vm to add such tracing - syslog, stdout - wherever, but somewhere.

Yes. I remember some mails you sent on this subject some months ago.

Mangled names are another topic, but I do NOT consider this to be a Squeak/Pharo/VM failing of any type. The author of the library failed to use extern "C" - simple as that.

I wonder why Python / Ruby can manage mangled names and that FFI don't. Because it's a real pain. Laurent.

The most recent encounter I had with this was on Windows, and I left behind the following comment:

The conspicuous (in hindsight) absence of extern "C" from the header file comes at a price: the names are mangled. Enter pexports.exe and some copy/paste with periodic acerbic remarks.

Translation: if you have not done this, consider using pexports.exe to obtain the true exported symbol names, and put them (with ?@, etc. intact) in your FFI definitions. HTH.

Bill

________________________________________ From: pharo-project-bounces@lists.gforge.inria.fr [ pharo-project-bounces@lists.gforge.inria.fr] On Behalf Of laurent laffont [laurent.laffont@gmail.com] Sent: Friday, May 06, 2011 2:34 PM To: Pharo-project@lists.gforge.inria.fr Subject: Re: [Pharo-project] Popularity of Smalltalk in Software Industry

On Fri, May 6, 2011 at 8:06 PM, Schwab,Wilhelm K <bschwab@anest.ufl.edu <mailto:bschwab@anest.ufl.edu>> wrote: No real agenda either way here, except to say that we almost have that in FFI. If you know enough to make the call, do you not know enough to write the FFI wrapper? Just asking. Having stirred the broth, I'll shut up and learn :)

My last attempts:

http://forum.world.st/Re-Z3950-on-OSX-module-yaz3-not-found-tp3161332p316133...

http://forum.world.st/Re-FFI-crashes-VM-on-Linux-was-Z3950-on-OSX-module-yaz...

Any help appreciated.

Laurent.

________________________________________ From: pharo-project-bounces@lists.gforge.inria.fr<mailto: pharo-project-bounces@lists.gforge.inria.fr> [ pharo-project-bounces@lists.gforge.inria.fr<mailto: pharo-project-bounces@lists.gforge.inria.fr>] On Behalf Of laurent laffont [laurent.laffont@gmail.com<mailto:laurent.laffont@gmail.com>] Sent: Friday, May 06, 2011 1:47 PM To: Pharo-project@lists.gforge.inria.fr<mailto: Pharo-project@lists.gforge.inria.fr> Subject: Re: [Pharo-project] Popularity of Smalltalk in Software Industry

On Fri, May 6, 2011 at 7:25 PM, Igor Stasenko <siguctua@gmail.com<mailto: siguctua@gmail.com><mailto:siguctua@gmail.com<mailto:siguctua@gmail.com>>> wrote: On 6 May 2011 18:45, laurent laffont <laurent.laffont@gmail.com<mailto: laurent.laffont@gmail.com><mailto:laurent.laffont@gmail.com<mailto: laurent.laffont@gmail.com>>> wrote:

...

On Fri, May 6, 2011 at 4:01 PM, Igor Stasenko <siguctua@gmail.com <mailto:siguctua@gmail.com><mailto:siguctua@gmail.com<mailto: siguctua@gmail.com>>> wrote:

...

Is it works because someone already provided a bindings for libraries you mention, or you did it by yourself?

Have just used ctypes:

...

...

...

import ctypes import ctypes.util ctypes.util.find_library("yaz") '/usr/local/lib/libyaz.dylib' yaz = ctypes.cdll.LoadLibrary("libyaz.dylib") con = yaz.ZOOM_connection_new("z3950.loc.gov<http://z3950.loc.gov><; http://z3950.loc.gov>", 7090);

oh cool.. give me a 5 minutes to load external library, look up the random symbol exported by it and make a call to it with random number of arguments, and expect it won't crash.

Works outta box! Cool.

Now serious. How about spending time writing a bindings for that library? Where you have properly organized classes/interfaces, error handling and other stuff. No takers?

Seriously, if you want such kind of shitty interface, which allows you to call arbitrary function with arbitrary arguments without any argument type checking, i can write it for you.

Do it, please do it. Seriously. Thanks

Laurent.

...

Laurent

-- Best regards, Igor Stasenko AKA sig.

On 6 May 2011 19:47, laurent laffont <laurent.laffont@gmail.com> wrote:

On Fri, May 6, 2011 at 7:25 PM, Igor Stasenko <siguctua@gmail.com> wrote:

...

On 6 May 2011 18:45, laurent laffont <laurent.laffont@gmail.com> wrote:

...

On Fri, May 6, 2011 at 4:01 PM, Igor Stasenko <siguctua@gmail.com> wrote:

...

Is it works because someone already provided a bindings for libraries you mention, or you did it by yourself?

Have just used ctypes:

...

...

...

import ctypes import ctypes.util ctypes.util.find_library("yaz") '/usr/local/lib/libyaz.dylib' yaz = ctypes.cdll.LoadLibrary("libyaz.dylib") con = yaz.ZOOM_connection_new("z3950.loc.gov", 7090);

oh cool.. give me a 5 minutes to load external library, look up the random symbol exported by it and make a call to it with random number of arguments, and expect it won't crash.

Works outta box! Cool.

Now serious. How about spending time writing a bindings for that library? Where you have properly organized classes/interfaces, error handling and other stuff. No takers?

Seriously, if you want such kind of shitty interface, which allows you to call arbitrary function with arbitrary arguments without any argument type checking, i can write it for you.

Do it, please do it. Seriously. Thanks Laurent.

Oh. Okay. I will put that in my years-long todo list. Remind me time to time :)

...

...

Laurent

-- Best regards, Igor Stasenko AKA sig.

-- Best regards, Igor Stasenko AKA sig.

On 6 May 2011 08:04, Jimmie Houchin <jlhouchin@gmail.com> wrote:

On 5/5/2011 8:19 PM, Igor Stasenko wrote:

...

On 6 May 2011 01:03, Jimmie Houchin<jlhouchin@gmail.com>  wrote:

...

On 5/5/2011 4:25 PM, Igor Stasenko wrote:

On 5 May 2011 21:39, Jimmie Houchin<jlhouchin@gmail.com>  wrote:

On 5/5/2011 12:21 PM, Igor Stasenko wrote:

On 5 May 2011 18:57, Jimmie Houchin<jlhouchin@gmail.com>    wrote:

[Big Snip]

Well said. Except that i'm not sharing your view that its hard to interface with foreign libraries. Its not hard at all. Of course to connect two different worlds, you need to have knowledge in both of them. But this requirement not a bit different when you take any other pair of languages and try to connect them.

Thanks,

I have no knowledge of either the knowledge or the challenges involved in using external libraries in Pharo or Squeak. I have no knowledge of FFI/Alien or using C/C++/C# or compilers.

So, read about it. It doesn't bites. :)

Yes, I know. But like many, my time is limited. I program my app in my spare time. I have a full-time+ job and a very large family who want some of my time.

No-no-no. Don't say that. We're all having own constraints. :)

...

...

As I said, I am totally unaware of the limitations of the system, but that it did what I need, and would have liked to do that from Pharo/Squeak. I know Dolphin has/had certain capabilities. But I don't prefer to use non-open source software if at all possible for development. I also am very preferential towards cross-platform software. Dolphin fails on all accounts. I would choose my Python/Pharo blend over Dolphin any day. I know it introduces some pain, but I am willing to accept the pain. I like tools that allow me to use them where ever I am and whatever I am doing.

However, that said, when I look at the facilities Python offers for such capabilities it on appearances looks quite impressive.

http://docs.python.org/library/ctypes.html

""" ctypes is a foreign function library for Python. It provides C compatible data types, and allows calling functions in DLLs or shared libraries. It can be used to wrap these libraries in pure Python.

15.18.1.1. Loading dynamic link libraries

ctypes exports the cdll, and on Windows windll and oledll objects, for loading dynamic link libraries.

You load libraries by accessing them as attributes of these objects. cdll loads libraries which export functions using the standard cdecl calling convention, while windll libraries call functions using the stdcall calling convention. oledll also uses the stdcall calling convention, and assumes the functions return a Windows HRESULT error code. The error code is used to automatically raise a WindowsError exception when the function call fails.

Here are some examples for Windows. Note that msvcrt is the MS standard C library containing most standard C functions, and uses the cdecl calling convention:

...

...

...

from ctypes import * print windll.kernel32 # doctest: +WINDOWS

<WinDLL 'kernel32', handle ... at ...>

...

...

...

print cdll.msvcrt # doctest: +WINDOWS

<CDLL 'msvcrt', handle ... at ...>

...

...

...

libc = cdll.msvcrt # doctest: +WINDOWS

So? Why do you think that you cannot do the same in Pharo? Have you digested what is available before doing it in Python?

I never said it couldn't be done in Pharo or Squeak. What I said is that I lack the skills to do so in Pharo.

It isn't a two lines of code an your off and running. Which is what Python provided me. Ok, in total, my Python code is more like 500 lines. It interfaces the COM DLL and serves the data to my Pharo app which provides all of my business logic. The Python app is simply an interface and gateway. No business logic.

Check the http://wiki.squeak.org/squeak/3773 It may not work with Pharo out of the box. But then it is really up to people who care about it: - do you want it to be working? Then do something, ask people, ask for help, organize & communicate.

...

...

""" It has examples for windows and linux.

I feel that you are seriously under-informed. Just try googling for:

Squeak + FFI Alien

NativeBoost

and just check what you can do using it:

http://www.squeaksource.com/NBOpenGL/

also check mail archives.

Forgive me, but i really can't understand , why you can't just look for what you need by yourself? Are google.com banned by your ISP?

For me this subject is interested in aspect, why this information (while being available openly) didn't catch your attention , so you using python instead. (Okay, you might miss some functionality like being able to automatically generate bindings to COM interfaces). But again, i am sure that if you surf the net or ask right questions on mailing list, you will probably discover that there is already work being done in that direction. And at last, if you feel that there is missing some key functionality which you want, and you expect that it is also could be useful to community, you can always implement it and share with people (instead of using Python ;). So, next who will come after you won't find himself in a desert with couple of oasises few hundred miles away.

Oh, I've Googled and searched and read the mailing lists. I've had off-list conversations with people who have used Alien/FFI about what I need. It is presently beyond my current skill and knowledge and time to acquire said knowledge. I am not saying it isn't doable. I am saying at the moment it isn't anything I have the time to pursue. I already have in place a working solution. It isn't as nice as  I would like, or an all Pharo solution. But it is working.

The current working solution is simply version 1.0 of my application. When it is up and running and I have further time I will explore other options.

Ok, as far mentioning this before here I go. :)

My messages

[snip]

As I stated, I am not a professional programmer. I have no computer science background. I have no C/C++/C# skills. The more time I spend doing low level stuff or learning how to do so, the less time I have to spend on what my app is intended to do, the business logic.

But you still need someone to do it.

My goal is more along the lines of getting my business up and running and hiring knowledgeable people like you to do the things that enable people like me.

I find Smalltalk enabling for the guy with an idea. Its the most enabling and empowering tool I've found for my ideas. I would like to see it improve in its enabling ability. I get excited about executing my ideas.

If I had no options but to learn Alien/FFI or ctypes, or C/C++/C#/F3, in order to build my application. I would. But it would delay me considerably. I have more ideas, than time.

So, find people who will help you. Apparently, if you decide to live with the status-quo and there is no any initiative from your side to improve the situation, then nothing will change. And then expect that next guy who will come here, will keep saying the same: yes i would love to use smalltalk, but i can't because it missing X, Y, Z.. so i will use Python instead.

I would like to see Pharo's story for interfacing external libraries to be better than Pythons. It would be very enabling for those of us with less skills, and less time for those who have the skills.

We're working on that. But i'd like to say that situation is not so different that in Python. Comparing to ctype , we have complete analog for it - FFI , Alien or NativeBoost

Thanks.

Jimmie

-- Best regards, Igor Stasenko AKA sig.

Exactly right. BUT! Very cool job. I'll have a look at it in detail once I start interfacing more thoroughly with real C apps outside of my own natives :) I wonder if being able to extend the object model of Smalltalk for native object wrappers would give you benefit... I guess it would? On 05/06/2011 10:30 AM, Stéphane Ducasse wrote:

Igor

We should face it: the FFI and Alien... are not that simple. Laurent spent time to try to cnnect to libs and it did not work. So we should really improve that aspect. Igor not everbody is able to code in assembler for nativeboost. So NB is probably a good insfrastructure but not ready for consumption for everybody. The interaction with C should be improved :)

Stef

...

So, read about it. It doesn't bites. :)

...

As I said, I am totally unaware of the limitations of the system, but that it did what I need, and would have liked to do that from Pharo/Squeak. I know Dolphin has/had certain capabilities. But I don't prefer to use non-open source software if at all possible for development. I also am very preferential towards cross-platform software. Dolphin fails on all accounts. I would choose my Python/Pharo blend over Dolphin any day. I know it introduces some pain, but I am willing to accept the pain. I like tools that allow me to use them where ever I am and whatever I am doing.

However, that said, when I look at the facilities Python offers for such capabilities it on appearances looks quite impressive.

http://docs.python.org/library/ctypes.html

""" ctypes is a foreign function library for Python. It provides C compatible data types, and allows calling functions in DLLs or shared libraries. It can be used to wrap these libraries in pure Python.

15.18.1.1. Loading dynamic link libraries

ctypes exports the cdll, and on Windows windll and oledll objects, for loading dynamic link libraries.

You load libraries by accessing them as attributes of these objects. cdll loads libraries which export functions using the standard cdecl calling convention, while windll libraries call functions using the stdcall calling convention. oledll also uses the stdcall calling convention, and assumes the functions return a Windows HRESULT error code. The error code is used to automatically raise a WindowsError exception when the function call fails.

Here are some examples for Windows. Note that msvcrt is the MS standard C library containing most standard C functions, and uses the cdecl calling convention:

...

...

...

from ctypes import * print windll.kernel32 # doctest: +WINDOWS <WinDLL 'kernel32', handle ... at ...> print cdll.msvcrt # doctest: +WINDOWS <CDLL 'msvcrt', handle ... at ...> libc = cdll.msvcrt # doctest: +WINDOWS So? Why do you think that you cannot do the same in Pharo? Have you digested what is available before doing it in Python?

...

"""

It has examples for windows and linux.

I feel that you are seriously under-informed. Just try googling for:

Squeak + FFI Alien

NativeBoost

and just check what you can do using it:

http://www.squeaksource.com/NBOpenGL/

also check mail archives.

Forgive me, but i really can't understand , why you can't just look for what you need by yourself? Are google.com banned by your ISP?

For me this subject is interested in aspect, why this information (while being available openly) didn't catch your attention , so you using python instead. (Okay, you might miss some functionality like being able to automatically generate bindings to COM interfaces). But again, i am sure that if you surf the net or ask right questions on mailing list, you will probably discover that there is already work being done in that direction. And at last, if you feel that there is missing some key functionality which you want, and you expect that it is also could be useful to community, you can always implement it and share with people (instead of using Python ;). So, next who will come after you won't find himself in a desert with couple of oasises few hundred miles away.

...

I didn't read it all as I am unqualified to assess its capabilities or limitations, nor do I presently need it as the win32 extensions do easily and well what I presently need to do.

Hopefully it can inform us as to what is expected and doable in alternative languages which are dynamic like Smalltalk.

Jimmie

-- Best regards, Igor Stasenko AKA sig.

Hi stefan I disagree with the C integration. This is still not really good in Pharo and I agree with Jimmie. Now for the documentation when did you send an help documentation for any part of the system? Or a bug fix? I find quite funny that people always talk but few are doing. We welcome comments/examples help. Stef

...

Igor

We should face it: the FFI and Alien... are not that simple. Laurent spent time to try to cnnect to libs and it did not work. So we should really improve that aspect. Igor not everbody is able to code in assembler for nativeboost. So NB is probably a good insfrastructure but not ready for consumption for everybody. The interaction with C should be improved :)

Stef

No, Stef, you didn't get it! It is documentation which needs to be improved in the first place. Documentation and its presentation. You know another reason why people love PHP? It is an ugly and awful language, but absolutely everything you ever want to do is covered in the docs, because people actually add examples and comments.

And frankly, you book efforts are certainly interesting, but far from being as visible and as accessible as online docs of other languages. That's what Smalltalk sucks at, accessible online documentation. People do google when they have a problem. If it does not pop up on the first page, it does not exists.

My impression of Smalltalk documentation is, it is non-existing. I usually do not even care to google, because it is unlikely that I find something. Instead I ask people. (And if you look at the traffic on the mailing list, thats not just me) That's a community issue, a problem of culture. I would never think about asking anyone if I have a problem with Python or PHP. I know everything is just a few google hits away.

Ah, and tooling of course...

-- Stefan Marr Software Languages Lab Vrije Universiteit Brussel Pleinlaan 2 / B-1050 Brussels / Belgium http://soft.vub.ac.be/~smarr Phone: +32 2 629 2974 Fax: +32 2 629 3525

...

Now for the documentation when did you send an help documentation for any part of the system? Or a bug fix? I find quite funny that people always talk but few are doing. We welcome comments/examples help. No need to get into a cat-fight here :)

No this is not my point. But what do people really do to help?

I do agree with Stefan, as well as with you that we could help; but I'm unaware of a good documentation effort outside of the image to start with. Camillo's website project might be the best effort I've seen until now; but then I didn't really look either. Maybe I missed some great website?

If this is just to spit out class comment on html I do not call that a documentation. Now we can take the book contents and generate html We have 350 pages in the first book and the same in the second one. People are free to join and write one or two chapters. Stef

On May 6, 2011, at 05:25, Toon Verwaest wrote:

...

If this is just to spit out class comment on html I do not call that a documentation. Now we can take the book contents and generate html We have 350 pages in the first book and the same in the second one. People are free to join and write one or two chapters. Generating HTML already seems like a good idea. I like the HTML version of SICP a lot for example.

+100 Google (and others) *are* key. I and many others have huge sets of bookmarks we never access because Google has become fast enough and authoritative enough that it's always the source I turn to first for information about almost anything. Turn the book into a web-based tutorial, and get all of us to link to it, get it listed in references for programming language stuff, such as Wikipedia pages on smalltalk etc, LtU, and others and do some other SEO. ../Dave

On 06 May 2011, at 11:17, Stéphane Ducasse wrote:

...

...

Now for the documentation when did you send an help documentation for any part of the system? Or a bug fix? I find quite funny that people always talk but few are doing. We welcome comments/examples help. No need to get into a cat-fight here :)

No this is not my point. But what do people really do to help?

...

I do agree with Stefan, as well as with you that we could help; but I'm unaware of a good documentation effort outside of the image to start with. Camillo's website project might be the best effort I've seen until now; but then I didn't really look either. Maybe I missed some great website?

If this is just to spit out class comment on html I do not call that a documentation. Now we can take the book contents and generate html We have 350 pages in the first book and the same in the second one. People are free to join and write one or two chapters.

Stéphane, Of course you are right: wining/complaining doesn't help, only action does. We need more people like Laurent focusing on documentation. I for one think that the well written, high quality books that exist (PBE, Seaside Book) are *very valuable*, much better than confusing wiki site (although these have their place as well). But the other point is: in most other popular languages, what current, young developers do, when they get an error that they don't understand is copy/paste the literal text in Google and in a surprisingly large number of cases you find some real answers in the first page. This is also related to popularity of course. Sven

On Fri, May 6, 2011 at 11:34 AM, Stéphane Ducasse <stephane.ducasse@inria.fr

wrote:

Probably we should find a way to generate HTML from PDF or Latex.

Any ideas?

I've used latex2html several years ago. +1 for html Laurent.

...

...

...

...

Now for the documentation when did you send an help documentation for any part of the system? Or a bug fix? I find quite funny that people always talk but few are doing. We welcome comments/examples help. No need to get into a cat-fight here :)

No this is not my point. But what do people really do to help?

...

I do agree with Stefan, as well as with you that we could help; but I'm unaware of a good documentation effort outside of the image to start with. Camillo's website project might be the best effort I've seen until now; but then I didn't really look either. Maybe I missed some great website?

If this is just to spit out class comment on html I do not call that a documentation. Now we can take the book contents and generate html We have 350 pages in the first book and the same in the second one. People are free to join and write one or two chapters.

Stéphane,

Of course you are right: wining/complaining doesn't help, only action does. We need more people like Laurent focusing on documentation.

I for one think that the well written, high quality books that exist (PBE, Seaside Book) are *very valuable*, much better than confusing wiki site (although these have their place as well).

But the other point is: in most other popular languages, what current, young developers do, when they get an error that they don't understand is copy/paste the literal text in Google and in a surprisingly large number of cases you find some real answers in the first page.

This is also related to popularity of course.

Sven

Guys, this is crazy, this is a runaway thread ;) Please, please, a bit more empathy for newcomers would not hurt: change the subject in such cases. Welcome to Sourav, don't be afraid to ask this list for anything obscure. and congratulations to Toon for initial answer. Nicolas 2011/5/6 Stéphane Ducasse <stephane.ducasse@inria.fr>:

Ok if some good souls want to give a try to generate html.

Stef

...

Probably we should find a way to generate HTML from PDF or Latex. Any ideas?

I've used latex2html several years ago.

+1 for html

Laurent.

...

...

...

...

Now for the documentation when did you send an help documentation for any part of the system? Or a bug fix? I find quite funny that people always talk but few are doing. We welcome comments/examples help. No need to get into a cat-fight here :)

No this is not my point. But what do people really do to help?

...

I do agree with Stefan, as well as with you that we could help; but I'm unaware of a good documentation effort outside of the image to start with. Camillo's website project might be the best effort I've seen until now; but then I didn't really look either. Maybe I missed some great website?

If this is just to spit out class comment on html I do not call that a documentation. Now we can take the book contents and generate html We have 350 pages in the first book and the same in the second one. People are free to join and write one or two chapters.

Stéphane,

Of course you are right: wining/complaining doesn't help, only action does. We need more people like Laurent focusing on documentation.

I for one think that the well written, high quality books that exist (PBE, Seaside Book) are *very valuable*, much better than confusing wiki site (although these have their place as well).

But the other point is: in most other popular languages, what current, young developers do, when they get an error that they don't understand is copy/paste the literal text in Google and in a surprisingly large number of cases you find some real answers in the first page.

This is also related to popularity of course.

Sven

Em 06/05/2011 06:34, Stéphane Ducasse < stephane.ducasse@inria.fr > escreveu:

Probably we should find a way to generate HTML from PDF or Latex.

From LaTeX to HTML obviously the path is to use LaTeX2HTML http://www.latex2html.org/ or if your LaTeX installation supports it htlatex to process the LaTeX sources into HTML pages. An older but still useful source of information is

http://mirror.softwarelivre.ufsc.br/pub/ctan/info/webguide/webguide.html#QQ1...

Any ideas?

Create some classes in Pharo to do the job?

Sorry for the flam below. I didn't have the time and energy to rewrite it.

...

...

Now for the documentation when did you send an help documentation for any part of the system? Or a bug fix? I find quite funny that people always talk but few are doing. We welcome comments/examples help. No, I report things on the mailing list, and I complain here and there about the narrow-mindedness of Smalltalk-evangelists. And when it comes to documentation, I document that RoarVM you might have heard of. If that is not interesting for you, sorry, we just don't share the same interests.

...

No need to get into a cat-fight here :)

No this is not my point. But what do people really do to help? Stef, if you haven't noticed: I don't care about Smalltalk, and I don't care about Pharo, or any other language out there in particular. I don't share your vision, I have other goals in life.

The only thing I do here is to point out the obvious (at least from my narrow-minded perspective).

If this is just to spit out class comment on html I do not call that a documentation. Now we can take the book contents and generate html We have 350 pages in the first book and the same in the second one. People are free to join and write one or two chapters.

That's what I mean. From my perspective, books about programming languages are a wast of effort. You need a good entry level book, that is all it takes. The rest is great online documentation. Unfortunately, Smalltalkers don't know anything outside the image... I don't know how other people work, but I never look into books when I program. They just don't work. They are slow, outdated and hard to search in. Honestly, I don't understand why books are a priority for you when you want to develop a community. You got a good entry level book, so what is the motivation to write another one? As I said, that is obviously my point of view, based on the way my workflow works. I prepare now Clojure assignments for my students, and there is also not a lot documentation out there, but all I need is centrally accessible at a place I easily identified with google. There is exactly one important window on my screen: http://clojure.github.com/clojure/clojure.core-api.html Thanks to my browsers search, everything is there. The Clojure book on my desk is just lying there and collecting dust... Best regards Stefan -- Stefan Marr Software Languages Lab Vrije Universiteit Brussel Pleinlaan 2 / B-1050 Brussels / Belgium http://soft.vub.ac.be/~smarr Phone: +32 2 629 2974 Fax: +32 2 629 3525

On 06 May 2011, at 13:56, Johan Brichau wrote:

When I program in Smalltalk, I have a lot of open code browsers in my image When I program in Java/Javascript/..., I have a lot of open web pages on my desktop

+1111 Browsing in a Smalltalk image should be self-documenting, not just because of comments, but because of beautiful, understandable code. A Dan Ingalls quote from "Design Principles Behind Smalltalk" ( http://www.cs.virginia.edu/~evans/cs655/readings/smalltalk.html ) Just to get warmed up, I'll start with a principle that is more social than technical and that is largely responsible for the particular bias of the Smalltalk project: Personal Mastery: If a system is to serve the creative spirit, it must be entirely comprehensible to a single individual. An excellent read, this document, by the way! Sven

El vie, 06-05-2011 a las 12:57 +0200, Stefan Marr escribió:

Sorry for the flam below. I didn't have the time and energy to rewrite it.

...

...

...

Now for the documentation when did you send an help documentation for any part of the system? Or a bug fix? I find quite funny that people always talk but few are doing. We welcome comments/examples help. No, I report things on the mailing list, and I complain here and there about the narrow-mindedness of Smalltalk-evangelists. And when it comes to documentation, I document that RoarVM you might have heard of. If that is not interesting for you, sorry, we just don't share the same interests.

...

...

No need to get into a cat-fight here :)

No this is not my point. But what do people really do to help? Stef, if you haven't noticed: I don't care about Smalltalk, and I don't care about Pharo, or any other language out there in particular. I don't share your vision, I have other goals in life.

So what are you doing here, the less the noise the best for everyone else. You don't contribute, you only criticize, and then you said you don't care, the we don't care about your "opinion". You can keep it for yourself.

The only thing I do here is to point out the obvious (at least from my narrow-minded perspective).

...

If this is just to spit out class comment on html I do not call that a documentation. Now we can take the book contents and generate html We have 350 pages in the first book and the same in the second one. People are free to join and write one or two chapters.

That's what I mean. From my perspective, books about programming languages are a wast of effort. You need a good entry level book, that is all it takes. The rest is great online documentation.

Umm, or you are very young or completely lost your mind. Internet as we know today has a decade at most. The documentation you as happy use now didn't existe 20 years ago. But, doesn't a lot of code was written more than 20 years ago. How is that possible just with fucking books and without google. That is BS.

Unfortunately, Smalltalkers don't know anything outside the image...

I don't know how other people work, but I never look into books when I program. They just don't work. They are slow, outdated and hard to search in.

Honestly, I don't understand why books are a priority for you when you want to develop a community. You got a good entry level book, so what is the motivation to write another one?

Because not everyone is so "smart" as you and we sometimes need to read a book in a more directed way that a simple API list and no way to see how is used. Also, check the Squeak wiki. Almost useless and nobody that is putting new content to it these days. On the other hands, the Pharo by example and seaside pier books or the HPI Seaside books are doing more for smalltalk than you want to accept. -- Miguel Cobá http://twitter.com/MiguelCobaMtz http://miguel.leugim.com.mx

(And some more flame war here) On 06 May 2011, at 17:11, Miguel Cobá wrote:

...

...

...

No need to get into a cat-fight here :)

No this is not my point. But what do people really do to help? Stef, if you haven't noticed: I don't care about Smalltalk, and I don't care about Pharo, or any other language out there in particular. I don't share your vision, I have other goals in life.

So what are you doing here, the less the noise the best for everyone else. You don't contribute, you only criticize, and then you said you don't care, the we don't care about your "opinion". You can keep it for yourself.

Well, I think my work on the RoarVM is some contribution, no? Perhaps, I would be more interested in Pharo if it would actually run nicely on the RoarVM, but I am stuck with a Squeak 3.x MVC image for my day to day work. And without anyone from the community approaching the work to make Pharo thread-safe that won't change. It is nice to change the world with Pharo, but the future is multi/manycore and Pharo does not support it. Ah, and the day has just 24h so don't expect anything from me beside the VM work, thats already enough to keep a whole team busy.

...

That's what I mean. From my perspective, books about programming languages are a wast of effort. You need a good entry level book, that is all it takes. The rest is great online documentation.

Umm, or you are very young or completely lost your mind. Internet as we know today has a decade at most. The documentation you as happy use now didn't existe 20 years ago. But, doesn't a lot of code was written more than 20 years ago. How is that possible just with fucking books and without google.

That is BS. How often do you see your colleagues going to the library, or the shelf next door to grab a book while they are programming? It has been 10 years ago that they stopped doing so.

Today, they google. Well, at least if they use another language than Smalltalk. And I don't think there is something inherently wrong with that.

to read a book in a more directed way that a simple API list and no way to see how is used. All I say is that a classical book is not the state of the art in the regard. Well written API documentation is. And well written API documentation includes examples. And, yes, Smalltalk people know about that.

I think, books are for introduction, but not to actually get work done. I need the knowledge directly at my finger tips.

On the other hands, the Pharo by example and seaside pier books or the HPI Seaside books are doing more for smalltalk than you want to accept.

Yes, they introduce people to Smalltalk. But this particular discussion started after someone mentioned that he/she used python since it was easier to bridge between Pharo and Python than it was to go from Pharo to the C code directly. That person knows Smalltalk, and is productive with it. Except for the things where he/she did not immediately found good documentation. That is the scenario I am talking about. And if I remember correctly there is even an article/book in the collection of free books on Stef's webpage that tells you how to implement primitives which could have done the job. But the problem here, he/she was just not able to find it. That's the point, and that is the problem I try to emphasize. It is certainly good to attract new people to Smalltalk, but how does that help if you can't make the ones who already know it to stay with it? With books? I think, Stef's efforts of writing books should be supported, but the content (if it is an open book) needs to be findable on Google and Bing... And that is not everything, the text has actually be in a form that it is easily accessible on the web. Which means, links, and repetition(for some degree of redundancy). (That is also what good API documentation does) Best regards Stefan -- Stefan Marr Software Languages Lab Vrije Universiteit Brussel Pleinlaan 2 / B-1050 Brussels / Belgium http://soft.vub.ac.be/~smarr Phone: +32 2 629 2974 Fax: +32 2 629 3525

(and here we go again...) On 06 May 2011, at 18:55, Miguel Cobá wrote:

El vie, 06-05-2011 a las 18:32 +0200, Stefan Marr escribió:

...

Well, I think my work on the RoarVM is some contribution, no? Perhaps, I would be more interested in Pharo if it would actually run nicely on the RoarVM, but I am stuck with a Squeak 3.x MVC image for my day to day work. And without anyone from the community approaching the work to make Pharo thread-safe that won't change. It is nice to change the world with Pharo, but the future is multi/manycore and Pharo does not support it. Ah, and the day has just 24h so don't expect anything from me beside the VM work, thats already enough to keep a whole team busy.

I found your post very contradicting and without internal consistency. You don't care about smalltalk but are creating a vm should run smalltalk (squeak or pharo) in multiple core. Don't get it. Or you care that you dedicate time to it or you don't care and don't know why you build a multicore vm for a system you don't care (maybe the money, the papers, the citations, don't know)

I am interested in VMs, so why do I need to care about the language on top? Actually, I do research in how to support all kind of different languages on top of the same VM, because there is not a single language that is the ultimate answer to all problems. That is why I do not care about any particular language. Just look at the JVM. How much of its technology was developed in the pure Java context? Not a lot. Most was actually conceived for Smalltalk. As long as the languages have some commonalities and are not based on graph-reduction like Haskell, then they usually don't require a completely new designed VM, but a nice set of common abstractions. So why, as a researcher, should I care about Smalltalk? Smalltalk is not the final answer, and will never be. Neither is any other single language.

...

...

...

That's what I mean. From my perspective, books about programming languages are a wast of effort. That is BS. How often do you see your colleagues going to the library, or the shelf next door to grab a book while they are programming? It has been 10 years ago that they stopped doing so.

That is the reason that O'Reilly books is broken and stop printing books 5 year ago right. Also this other publisher, I think they use to call themselves the pragmatic progammer, what a bunch of loser, trying to publish new books in this days that google has everything that is to know.

Wait, no, they exists and publish books, maybe is for having something to do with their free time. Just answer the question! How often do your colleagues grab a book from the shelf? How often a day, how often a week?

...

Today, they google. Well, at least if they use another language than Smalltalk. And I don't think there is something inherently wrong with that.

Yes but isn't the only way to get knowledge. No, but to my experience books can only be a help to get an initial overview. How often do you come across a book that covers a topic in such a depth that you actually can use it when you have a specific question while working?

Especially the Pragmatic Programmer books you refer to, most of them talk about the basic concepts. They are not full references. Nice introductions, but nothing to actually work with when you solve a deeper going problem.

Maybe assembler should be lectured no more, because we have high level programing languages and nobody needs that old tech anymore! What are you talking about?

You said:

...

...

That's what I mean. From my perspective, books about programming languages are a wast of effort.

and then:

...

I think, Stef's efforts of writing books should be supported,

inconsistent! Right, it is indeed inconsistent, since that statement now also includes having books for introductory purposes. But again, that was not the initial point of the discussion.

Still, there is already a good thing coming out of the whole flame war. Laurent took the time to look into how to make the books web-accessible. All it took was a small nudge in the right direction. Best regards Stefan -- Stefan Marr Software Languages Lab Vrije Universiteit Brussel Pleinlaan 2 / B-1050 Brussels / Belgium http://soft.vub.ac.be/~smarr Phone: +32 2 629 2974 Fax: +32 2 629 3525

On 06/05/2011 22:23, Stefan Marr wrote:

(and here we go again...)

On 06 May 2011, at 18:55, Miguel Cobá wrote:

...

El vie, 06-05-2011 a las 18:32 +0200, Stefan Marr escribió:

...

Well, I think my work on the RoarVM is some contribution, no? Perhaps, I would be more interested in Pharo if it would actually run nicely on the RoarVM, but I am stuck with a Squeak 3.x MVC image for my day to day work. And without anyone from the community approaching the work to make Pharo thread-safe that won't change. It is nice to change the world with Pharo, but the future is multi/manycore and Pharo does not support it. Ah, and the day has just 24h so don't expect anything from me beside the VM work, thats already enough to keep a whole team busy.

I found your post very contradicting and without internal consistency. You don't care about smalltalk but are creating a vm should run smalltalk (squeak or pharo) in multiple core. Don't get it. Or you care that you dedicate time to it or you don't care and don't know why you build a multicore vm for a system you don't care (maybe the money, the papers, the citations, don't know) I am interested in VMs, so why do I need to care about the language on top? Actually, I do research in how to support all kind of different languages on top of the same VM, because there is not a single language that is the ultimate answer to all problems. That is why I do not care about any particular language. Just look at the JVM. How much of its technology was developed in the pure Java context? Not a lot. Most was actually conceived for Smalltalk. As long as the languages have some commonalities and are not based on graph-reduction like Haskell, then they usually don't require a completely new designed VM, but a nice set of common abstractions. So why, as a researcher, should I care about Smalltalk? Smalltalk is not the final answer, and will never be. Neither is any other single language.

Smalltalk the language no. Smalltalk the infinitely malleable environment? That's another question :)

...we will convert our lovely books from pdf to html-crap.

Please do keep the PDFs crawlable by search engines, though. I get PDF hits in my search results all the time, and I appreciate being able to see the layout the way the authors intended. My PDF viewer can search through the text just fine, too. thanks again for the books! -C -- Craig Latta www.netjam.org/resume +31 06 2757 7177 + 1 415 287 3547

we will ;) Don't worry :) On May 7, 2011, at 3:29 PM, Craig Latta wrote:

...

...we will convert our lovely books from pdf to html-crap.

Please do keep the PDFs crawlable by search engines, though. I get PDF hits in my search results all the time, and I appreciate being able to see the layout the way the authors intended. My PDF viewer can search through the text just fine, too.

thanks again for the books!

-C

-- Craig Latta www.netjam.org/resume +31 06 2757 7177 + 1 415 287 3547

On 06 May 2011, at 19:08, Igor Stasenko wrote:

I think the main reason why RoarVM does not exists for Pharo is because there was no discussion and planning beforehead, how we could cooperate. Where the discussion, how we could introduce new execution models, and gradually (means step by step) migrate to new VM?

Igor, I remember some discussion from the Smalltalk school, two important points: 1. it is not a new execution model, but you do not have your old scheduling guarantees anymore. 2. "all it takes" is to make the important libraries thread-safe There is nothing fundamental in the RoarVM that is changing the language semantics of Smalltalk. It is just that for: `[do something] fork` you will have to assume that it is executed in parallel to other code.

Take into account there there is virtually no knowledge outside of your team, what has to be changed in order to make Pharo run on RoarVM. So what did you expected? That people drop everything which already works well for them and hastily migrate to new platform? No, of course not. I never expected anyone to jump onto a research platform, and it is also not really about adopting our C++ code, but about trying to integrate the ideas into the standard VM with Cog.

Talking with Eliot basically resulted in the insight that there is no major problem that would prevent the adoption of parallel execution in the CogVM. They only thing we became aware of that might be interesting are the PICs. And there he had a nice paper reference with a perfectly good solution...

How about CogVM? Should we stop developing it? Or we should start supporting both? And can we do that without too much pain? Give us the idea. It is all about adopting the ideas, and I could collaboration on that, but I can't do that work. The only problem I see is that there does not seem to be a business case for a CogVM with parallel Smalltalk Process execution.

But then i wouldn't come to pharo list to say that i don't give a shit about pharo. Igor, the fact that the community did not show active interest in the RoarVM is much older than my bold statement. And, please understand my statement in the context of me not giving a damn about any particular language. Seriously, I can stand those 'ah but my language is so much nicer' discussions. And this here is just a variation of it. Languages are tools, and they are useful as long as they fit the problem.

I appreciate the engineering effort what you have did. But its a top of the iceberg. Migrating existing system to another platform is very complex task,

Actually, for the RoarVM it is not. It is a lot of work, but it is not complex. It is tedious, but it can be done in small well defined steps. Perfect small projects for interested students. I already convinced two guys to apply for the Google Summer of Code. But well, that never happened... So, I do what I can, and actually more. But the community only responds to me when I say that I hate Smalltalk. Strange, isn't it? Best regards Stefan -- Stefan Marr Software Languages Lab Vrije Universiteit Brussel Pleinlaan 2 / B-1050 Brussels / Belgium http://soft.vub.ac.be/~smarr Phone: +32 2 629 2974 Fax: +32 2 629 3525

On 07 May 2011, at 02:06, Igor Stasenko wrote:

On 6 May 2011 23:45, Stefan Marr <pharo@stefan-marr.de> wrote:

...

There is nothing fundamental in the RoarVM that is changing the language semantics of Smalltalk.

It is just that for: `[do something] fork` you will have to assume that it is executed in parallel to other code.

only that? Heh.. then we're 99% done. Except that this last 1% is still could take years to complete :)

But why? We show with the Squeak 3.x MVC image that it is perfectly possible. (the change set is absolutely minimal, and the only thing we needed to fix was some Delay related issues) And, on top of that it is perfectly possible, there is a smooth transition path. You can tell all your Process objects to just run on the main core in the beginning, and then just use the other cores for code you wrote specifically. Thus, there is a nice step-wise path. And I think I outlined that in my 20min presentation at the school after Stef asked. BTW: are those videos somewhere accessible?

From my perspective all it takes is someone who actually cares, and has the time to experiment a bit with parallel programming.

The kernel, and the tools can be migrated step by step. However, that is just on top of the current RoarVM, which is certainly not as attractive as a CogVM with real thread support.

I can tell you more: there is no business cases for VM(s) which can do manycore :) No manycore perhaps, but luckily the low-power end is pushing to multicore solutions. So, your next iPhone and Android app will need to leverage at least two cores for optimal performance, and even quad-core phones are just around the corner.

Implementing VM which enabling massive parallelism is just a beginning. Then obviously you need do to a lot at language side to leverage that, in order to really say "yes, our system(s) are aware of multicore and can scale almost linearly in future". No, don't think so. If people got deadlines and a need, they will do it with the tools at hand. And what I am doing here is absolutely not rocket science. Actually it is pretty hard to sell the engineering we do as science at all. Because the multicore VM problem has been solved a decade ago. They just forgot to tell us how...

Best regards Stefan -- Stefan Marr Software Languages Lab Vrije Universiteit Brussel Pleinlaan 2 / B-1050 Brussels / Belgium http://soft.vub.ac.be/~smarr Phone: +32 2 629 2974 Fax: +32 2 629 3525

On 07 May 2011, at 09:19, Stéphane Ducasse wrote:

when and where did you announce that? Where is the code?

http://forum.world.st/RoarVM-The-Manycore-SqueakVM-td3025321.html And here the MVC compatibility code: https://github.com/smarr/RoarVM/raw/master/image.st/RVM-multicore-support.mv... And there is even a file for minimal Pharo compatibility: https://github.com/smarr/RoarVM/raw/master/image.st/RVM-multicore-support.ph... -- Stefan Marr Software Languages Lab Vrije Universiteit Brussel Pleinlaan 2 / B-1050 Brussels / Belgium http://soft.vub.ac.be/~smarr Phone: +32 2 629 2974 Fax: +32 2 629 3525

On 07 May 2011, at 10:41, Stéphane Ducasse wrote:

Too bad. The eclipse license is not for us. I know, other reason to 'steal' the ideas and integrate them into Cog. And I can't do that, because all code that is derived from what we do with the RoarVM will probably need to be EPL (as far as I understand, but I am not a lawyer...) However, I don't see a problem in guiding someone...

Now do you imply that Process should use your primitive?

The lengthy discussion is available here: (and in the mailing list archives) Multiprocessor Smalltalk: A Case Study of a Multiprocessor-Based Programming Environment by Joseph Pallas, and David Ungar, in Proceedings of the ACM SIGPLAN 1988 Conference on Programming Language Design and Implementation, ACM (1988), p. 268-277. http://portal.acm.org/citation.cfm?id=54017 The short version: There is exactly one important primitive in the VM: to determine the process that is locally executed. And the image needs to be adapted to get rid of the activeProcess semantic of being a singleton. Following Pallas, that is thisProcess in our code, and that needs to be used for getting the locally executing process. That requires some adaptations here and there like in the debugger. The other primitives are for greater flexibility. Best regards Stefan -- Stefan Marr Software Languages Lab Vrije Universiteit Brussel Pleinlaan 2 / B-1050 Brussels / Belgium http://soft.vub.ac.be/~smarr Phone: +32 2 629 2974 Fax: +32 2 629 3525

On 07 May 2011, at 16:33, Igor Stasenko wrote:

On 7 May 2011 10:41, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote:

...

Too bad. The eclipse license is not for us.

I'm not a lawyer, but reading it, it looks like this license are compatible with MIT.

My understanding is that the EPL requires 'derived work' to be EPL licensed. While the EPL has much clearer definition of derived work than the GPL, it is still not as free as the MIT license. (Derived work explicitly excludes software modules build on top, so that it is clear that your business code can stay closed source. That definition is missing in the GPL as far as I understand.) But I can't change that. The IBM lawyers insist on EPL. Anyway, the RoarVM will never be as performant as the CogVM, so, the only reasonable approach is to steal the ideas and bring them to Cog. Most ideas are described in papers already. And some missing details are hopefully published soon... -- Stefan Marr Software Languages Lab Vrije Universiteit Brussel Pleinlaan 2 / B-1050 Brussels / Belgium http://soft.vub.ac.be/~smarr Phone: +32 2 629 2974 Fax: +32 2 629 3525

On 7 May 2011 07:25, Michael Haupt <mhaupt@gmail.com> wrote:

Hi Igor,

Am 07.05.2011 um 02:06 schrieb Igor Stasenko <siguctua@gmail.com>:

Heh.. then we're 99% done. Except that this last 1% is still could take years to complete :)

you serious? Making a Smalltalk image thread-safe is that hard?

I can tell you more: there is no business cases for VM(s) which can do manycore :) At least, to my perception, there is not much pressure from people (even on mainstream languages) to leverage this technology.

What market are you talking about? There are lots and lots of customers of certain big database/appliance vendors that would loudly protest had the stuff they bought no inherent support for parallel execution.

So, where those customers to sponsor development of VM(s) they need so much?

I thought it is very close to come into our houses, but no.. it stuck somewhere on marketplace, buying a better clothes.

And now it's arrived in tablets already. It's your problem if you leave your iPad lying around *outside* your house. ;-)

I am not talking about hardware, i'm talking about software. It is largely not using parallelism, or using it wrong.

Implementing VM which enabling massive parallelism is just a beginning. Then obviously you need do to a lot at language side to leverage that, in order to really say "yes, our system(s) are aware of multicore and can scale almost linearly in future".

And do not forget the most important thing: people must be educated to program for parallelism. The computing power is there to be exploited, but programmers need to know how to exploit it. You can't rely on the compiler or VM alone.

That's the main problem, IMO. Most smalltalkers i met (oh.. why.. not just smalltalkers), fear parallelism like fire. Refer to it as something too complex they can't understand. And i think that's why "Parallelism to masses" approach is largely failing , not just in smalltalk, but in whole software industry.

Now, given that squeak's code a hugely non-thread-safe, imagine how many iterations in redesign it would take to make it thread-safe first, and then make it run in parallel with rest of the stuff.

Is it really *that* hard or is this just FUD?

Its not that hard. But if you take the previous paragraph into account...

Best, Michael

-- Best regards, Igor Stasenko AKA sig.

Hi Igor, Am 07.05.2011 um 13:37 schrieb Igor Stasenko <siguctua@gmail.com>:

...

What market are you talking about? There are lots and lots of customers of certain big database/appliance vendors that would loudly protest had the stuff they bought no inherent support for parallel execution.

So, where those customers to sponsor development of VM(s) they need so much?

for the market you probably have in mind, I do not know. Perhaps that market is not so interesting after all.

I am not talking about hardware, i'm talking about software. It is largely not using parallelism, or using it wrong.

That is precisely my point.

...

Is it really *that* hard or is this just FUD?

Its not that hard. But if you take the previous paragraph into account...

So it is about fear. Which in turn means that it would take some fearless individuals (not necessarily leaders). I'm not afraid of parallelism, and think it would be an interesting and worthwhile experiment. Stefan would probably agree, and he has some experience with an older Squeak image. I'm sure you agree. How about it, guys? ;-)

Best, Michael

I do agree that we could do something in this regard. But given that we could "do something" almost everywhere, it is hard to say if this topic should take priority. Currently my plate is full. So, its not about that there are someone says "no we don't want it ", its mainly about lack of resources. And in addition, the whole OOPSLA world is working on it already so ... why bother ;P

On 05/07/2011 07:27 PM, Michael Haupt wrote:

Toon,

Am 07.05.2011 um 16:56 schrieb Toon Verwaest<toon.verwaest@gmail.com>:

...

...

I do agree that we could do something in this regard. But given that we could "do something" almost everywhere, it is hard to say if this topic should take priority. Currently my plate is full. So, its not about that there are someone says "no we don't want it ", its mainly about lack of resources. And in addition, the whole OOPSLA world is working on it already so ... why bother ;P oh. I had not known the OOPSLA community was devoting so much time to improvements in certain Smalltalk implementations. ;-) Point well taken. I wasn't talking about Smalltalk of course :) Best,

Michael

On 07. 05. 2011 13:37, Igor Stasenko wrote:

On 7 May 2011 07:25, Michael Haupt <mhaupt@gmail.com> wrote:

...

And do not forget the most important thing: people must be educated to program for parallelism. The computing power is there to be exploited, but programmers need to know how to exploit it. You can't rely on the compiler or VM alone.

That's the main problem, IMO. Most smalltalkers i met (oh.. why.. not just smalltalkers), fear parallelism like fire. Refer to it as something too complex they can't understand. And i think that's why "Parallelism to masses" approach is largely failing , not just in smalltalk, but in whole software industry.

Web frameworks are ones who would benefit from multicore Smalltalk instantly and without exposing concurrent programing complexities to web app developers. All paralell programing/execution complexity can be nicely encapsulated/hidden in web framework itself. And actually is now already, because we need to handle parallel web requests all the time. Best regards Janko -- Janko Mivšek Aida/Web Smalltalk Web Application Server http://www.aidaweb.si

On 7 May 2011 08:54, Douglas Brebner <squeaklists@fang.demon.co.uk> wrote:

On 07/05/2011 01:06, Igor Stasenko wrote:

...

On 6 May 2011 23:45, Stefan Marr<pharo@stefan-marr.de>  wrote:

...

On 06 May 2011, at 19:08, Igor Stasenko wrote:

...

How about CogVM? Should we stop developing it? Or we should start supporting both? And can we do that without too much pain? Give us the idea.

It is all about adopting the ideas, and I could collaboration on that, but I can't do that work. The only problem I see is that there does not seem to be a business case for a CogVM with parallel Smalltalk Process execution.

I can tell you more: there is no business cases for VM(s) which can do manycore :) At least, to my perception, there is not much pressure from people (even on mainstream languages) to leverage this technology. I thought it is very close to come into our houses, but no.. it stuck somewhere on marketplace, buying a better clothes.

I have the bad feeling that there will continue to be no business case for manycore VMs (or serious concurrency in general) until we hit a wall at which point they'll suddenly go from uninteresting to desperately essential and any language not able to make the jump will be in big trouble.

Besides, wasn't part of Smalltalk all about leading the way rather than following behind?

To lead, you need to have plenty of resources at your disposal. Human and material ones. -- Best regards, Igor Stasenko AKA sig.

...

Ah, and the day has just 24h so don't expect anything from me beside the VM work, thats already enough to keep a whole team busy.

This statement is true for most of us, just replace "VM" with something else.

About RoarVM. I think the main reason why RoarVM does not exists for Pharo is because there was no discussion and planning beforehead, how we could cooperate. Where the discussion, how we could introduce new execution models, and gradually (means step by step) migrate to new VM?

+1

Take into account there there is virtually no knowledge outside of your team, what has to be changed in order to make Pharo run on RoarVM. So what did you expected? That people drop everything which already works well for them and hastily migrate to new platform?

How about CogVM? Should we stop developing it? Or we should start supporting both? And can we do that without too much pain? Give us the idea.

I, personally all for having VM which knows how to exploit multicore/manycore systems. But at what costs?

+1

At cost of throwing away everything and implementing new VM and new smalltalk from scratch? I can do that myself. But then i wouldn't come to pharo list to say that i don't give a shit about pharo. Because if i would like Pharo to use my stuff, i will stay with people and help, and try to figure out how we can manage to leverage new technology without jumping too high with the risk to break the legs.

+1

It makes no point saying here that you don't give a shit about Pharo. If you don't, why pharoers should?

I appreciate the engineering effort what you have did. But its a top of the iceberg. Migrating existing system to another platform is very complex task, and given that day has 24 hours in it.. figure the answer :)

On Fri, May 6, 2011 at 4:05 PM, Toon Verwaest <toon.verwaest@gmail.com> wrote:

...

Now for the documentation when did you send an help documentation for any part of the system? Or a bug fix? I find quite funny that people always talk but few are doing. We welcome comments/examples help.

No need to get into a cat-fight here :)

I do agree with Stefan, as well as with you that we could help; but I'm unaware of a good documentation effort outside of the image to start with. Camillo's website project might be the best effort I've seen until now; but then I didn't really look either. Maybe I missed some great website?

You are welcome to join us on the documentation effort: - Pharo by Example vol 1 and vol 2 & translations : https://github.com/SquareBracketAssociates We are looking for more people to write chapters and translate them in other languages. - the collaboractive Pharo Book: http://book.pharo-project.org/ - participate to COTC (Comment Of The Day Contest) organized by Laurent: http://code.google.com/p/pharo/wiki/CommentOfTheDayContest in order to comments Pharo classes. This is not enough for you ;-) Regards, -- Serge Stinckwich UMI UMMISCO 209 (IRD/UPMC), Hanoi, Vietnam Every DSL ends up being Smalltalk http://doesnotunderstand.org/

...

Regards,

Something like the Pharo book that would contain also explanation about the language itself (and its libraries) would be great. Something like "teach yourself scheme in fixnum days" for Smalltalk. I loved the "teach yourself scheme in fixnum days"

may be you should open the pdf of pharo by example...... or may be not Stef

So what do you suggest? On 05/06/2011 10:55 AM, Stefan Marr wrote:

On 06 May 2011, at 10:30, Stéphane Ducasse wrote:

...

Igor

We should face it: the FFI and Alien... are not that simple. Laurent spent time to try to cnnect to libs and it did not work. So we should really improve that aspect. Igor not everbody is able to code in assembler for nativeboost. So NB is probably a good insfrastructure but not ready for consumption for everybody. The interaction with C should be improved :)

Stef No, Stef, you didn't get it! It is documentation which needs to be improved in the first place. Documentation and its presentation. You know another reason why people love PHP? It is an ugly and awful language, but absolutely everything you ever want to do is covered in the docs, because people actually add examples and comments.

And frankly, you book efforts are certainly interesting, but far from being as visible and as accessible as online docs of other languages. That's what Smalltalk sucks at, accessible online documentation. People do google when they have a problem. If it does not pop up on the first page, it does not exists.

My impression of Smalltalk documentation is, it is non-existing. I usually do not even care to google, because it is unlikely that I find something. Instead I ask people. (And if you look at the traffic on the mailing list, thats not just me) That's a community issue, a problem of culture. I would never think about asking anyone if I have a problem with Python or PHP. I know everything is just a few google hits away.

Ah, and tooling of course...

On 6 May 2011 10:30, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote:

Igor

We should face it: the FFI and Alien... are not that simple. Laurent spent time to try to cnnect to libs and it did not work. So we should really improve that aspect. Igor not everbody is able to code in assembler for nativeboost.

I am not presumed that everyone should. But just don't fall into delusion, that you can start using external library (either written in C or C# or in Java), without knowing at all the foreign language in which library are written, and just after few mouse clicks in Pharo image. The library are written not in smalltalk, and if you need to use it, you have to be prepared. You cannot avoid that. That's my point.

So NB is probably a good insfrastructure but not ready for consumption for everybody.

Mind telling me, what is missing? I am eager for improving and making it easy to use.

The interaction with C should be improved :)

How? Any clues? How else it could be improved in addition to be able to define, which function you going to call and using which library? How else it could be done? Parsing C headers and automatically generating bindings? This is doable. But it still will require manual effort. You cannot build an auto-generator to be intelligent enough to detect all specific aspects, which used in library. Read a http://www.swig.org/ they having a plugin system (and at some point i even written a smalltalk plugin for it). But even then, for generating a specific library bindings, you still have to provide a script/configuration for your library. Because if you run it in automatic mode, in most cases you'll end up with a lot of useless and/or broken stuff. Translating (producing wrappers) from one language into another requires intelligent effort, it cannot be fully automated.

Stef

-- Best regards, Igor Stasenko AKA sig.

On 6 May 2011 15:42, Denis Kudriashov <dionisiydk@gmail.com> wrote:

2011/5/6 Igor Stasenko <siguctua@gmail.com>

...

But just don't fall into delusion, that you can start using external library (either written in C or C# or in Java), without knowing at all the foreign language in which library are written, and just after few mouse clicks in Pharo image.

Hello,

In last week I implement java binding with some C library (it containes callbacks). Before I never doing this kind of work in java. And It will not simple task. I use JNA - analogous FFI libraries in smalltalk. So I done it. And It work from my tests. But when I deploy It to apatch server as servlet it cant work. Java craches in arbitrarily places of dll calling.Solutions will be simplest C dll wrapper which provides two functions without callbacks. Similar problems will be when we doing java connect to com-object. Solutions will be C# socket server :))

So Java connection to foreign world is not easy. Although there is good docs.

That's what i trying to say. Someone have to get his hands dirty and do it. Fail, fix & repeat until its done. There is no magic. And it always will be hard and error prone. -- Best regards, Igor Stasenko AKA sig.

Haha.. i have strong suspicion that here you are talking about quite specific set of libraries, which using OLE/COM interfaces. Indeed, one could implement an automatic "import/connect" tool for it, because a library itself contain enough information reflecting it interface(s). You can check Dolphin smalltalk which works only on windows and has integrated solution for that for years: In same way like you described, you just pick the library, click "ok" and its done & ready for use.

True, I was just ready to write the same thing about Dolphin and works very well, transforming OLE objects in Smalltalk objects with their respective methods. But allways you need to know what to do with such objects and methods. And is very dependent of the operating system and the underlying technology. I think that the cooperation with the outside world is needed, but I will not miss the advantages of work in Smalltalk, lot of times is more cheap to develop the needed stuff in Smalltalk than a complex artifact to interact with another technology. Cheers. Germán.

On 5 May 2011 17:32, Toon Verwaest <toon.verwaest@gmail.com> wrote:

On 05/05/2011 05:26 PM, Cédrick Béler wrote:

...

...

Lastly, most Smalltalk systems are image based...

...which makes you feel the system is "alive", hence one **huge benefit** of Smalltalk: its debugger which enables on the fly debbuging... and also test driven development (real one [1]) where you can run incomplete code and code what's missing iteratively when you need it (Smalltalk is a live system, not only a language as somebody said lately).

Cédrick

[1] see in particular this webcast: http://www.pharocasts.com/2010/01/starting-with-sunit-and-debugger.html

Seriously ... these points in favor of the image are so m00t. Lets see how it would work without an image:

I write a C application which I link to GCC. Now I run GDB on my application, and while running I have the whole GCC compiler collection at my disposal while running. While debugging (at some breakpoint) I just let the GCC library compile some C code for me; I turn on the executable flag and whooptidoo, I have a Smalltalk like debugger for C.

This is totally unrelated to having an image; it's just a great debugger implementation. 2 completely different things. No reason why this wouldn't work for C; except for the fact that they didn't do it yet (those lazy bastards).

Yes, but it affects the way how you design your applications and frameworks. In image-based setup you tend to do things in a way, that things should be always available at any moment (wanna create an instance of me? - feel free to do it). In C and other languages which having compile & run cycle it is totally different. Things are available for you not instantly, but only after you restart everything from scratch, reinitialize everything and then repeat exactly same steps to meet the same point as before. This is often tedious, time consuming and frustrating. And this difference in approaches leads to fatal flaws in design, like on Windows OS, which requires you to reboot machine every time you install new driver(s). -- Best regards, Igor Stasenko AKA sig.

2011/5/5 Cédrick Béler <cdrick65@gmail.com>:

Lastly, most Smalltalk systems are image based...

...which makes you feel the system is "alive", hence one **huge benefit** of Smalltalk: its debugger which enables on the fly debbuging... and also test driven development (real one [1]) where you can run incomplete code and code what's missing iteratively when you need it (Smalltalk is a live system, not only a language as somebody said lately).

Super Full Agree!

Cédrick [1] see in particular this webcast: http://www.pharocasts.com/2010/01/starting-with-sunit-and-debugger.html

--