- Pharo-dev - lists.pharo.org

[Pharo-project] When is a freeze a bug or me fiddling too much
by Kasper Ãsterbye Dec. 16, 2011

Dec. 16, 2011

Hi all, I have spend some time fiddling with Pharo, wonderful! I have always had a tendency to redefine vital methods in ST, and sometimes that kills it (typically it just hangs). Pharo seems a little more prone to that than what I was used to. In particular, I seem to have a tendency to spot methods where inserting a "self halt" will cause the pre-debugger to open a new pre-debugger to open a new pre-debugger to.... My question is - are there any guidelines on when I should just live with it, or if it is a bug? Would there be anyone interested in me collecting a list of the things that I stumble upon that make the image freeze? Best, Kasper

4 4

[Pharo-project] [update 1.4] #14261
by Marcus Denker Dec. 16, 2011

Dec. 16, 2011

14261 ----- Issue 4745: Debugger opens when trying to get unimplemented calls http://code.google.com/p/pharo/issues/detail?id=4745 Issue 4870: numargs? http://code.google.com/p/pharo/issues/detail?id=4870 Issue 5111: Make TestRunner search input ignore case http://code.google.com/p/pharo/issues/detail?id=5111 Issue 5087: remove unsent methods from ServerDirectory http://code.google.com/p/pharo/issues/detail?id=5087 -- Marcus Denker -- http://marcusdenker.de

1 0

[Pharo-project] [update 1.4] #14260
by Marcus Denker Dec. 16, 2011

Dec. 16, 2011

14260 ----- Issue 5084: Remove textStyle FileOut http://code.google.com/p/pharo/issues/detail?id=5084 Issue 5108: InflateStream bug with empty strings http://code.google.com/p/pharo/issues/detail?id=5108 Issue 4920: Copying a trait fails http://code.google.com/p/pharo/issues/detail?id=4920 Issue 5081: Remve Morph saving to file/loading http://code.google.com/p/pharo/issues/detail?id=5081 -- Marcus Denker -- http://marcusdenker.de

1 0

[Pharo-project] FFI is broken in Cog [WAS] Re: Pharo 1.3 vs 1.2.1 some results for ConfigurationOfODBC - success
by Mariano Martinez Peck Dec. 16, 2011

Dec. 16, 2011

On Wed, Dec 14, 2011 at 5:10 PM, Ben Coman <btc(a)openinworld.com> wrote: > ** > > That works great. Thanks Mariano. Since I got the file direct from you I > have uploaded it to [2] in case it is of use to others. > > I see the file you provide is "Pharo-1.3-13307-OneClickWin.zip" has many > more dlls than the file of the same name from INRIA>Files [1]. Is this a > fix that can be added by a user to the released Pharo-1.3-13315, or > otherwise be flagged to go into a possible Pharo-1.3.1 ? > > I was exploring the load sequence and it was interesting to discover that > in a fresh image, after... > > Gofer new > > squeaksource: 'ODBC'; > > package: 'ConfigurationOfODBC'; load. > > (Smalltalk at: #ConfigurationOfODBC) load. > > ..."listLoadedModules" did not show odbc32. This only appeared after > executing... > > connection := ODBCConnection dsn: 'TestMDB' user: '' password: ''. > > That is cool seeing the dynamic nature of plugin loading. Now while you > will already be aware of the difference between your supplied file [2] and > the one from [1], for my own learning attempt it would seem from > "SmalltalkImage current listLoadedModules" that the difference is... > SqueakFFIPrims VMMaker-oscog.54 (e) working > versus > SqueakFFIPrims VMMaker-oscog-IgorStasenko.123 (e) not working. > ... and unfortunately there is no version information on odbc32. > Yes, the zip I provided you is a CogVM one click I did myself for windows because the default is not working. It was always a mess for me. I didn't even remember what I did to make it work. I think I took the SqueakFFI dll from an older Cog VM or from the one build by Jenkins or something like that. Until I make it work. But out of the box, the FFI is not working. Maybe it has been fixed. You can try the very last VM of Eliot (since there is no Jenkins for Windows yet) http://www.mirandabanda.org/files/Cog/VM/VM.r2522/cogwin.zip is it working? > > A side comment for whoever maintains ConfigurationOfODBC... I wonder if > this could be made to fail earlier - at the "(Smalltalk at: > #ConfigurationOfODBC) load" rather than waiting for the first attempt at a > connection. I think logically proving access to the external library > should be part of the package load. This would help isolate such issues for > users new to the package. I notice that after a SaveAndExit and restart, > the odbc plugin is no longer shown in "listLoadedModules" until dynamically > reloaded on next use of "ODBCconnection". This would remain the standard > behaviour after proving access to the library for the first time during the > package load. > +1 > > [1] https://gforge.inria.fr/frs/?group_id=1299&release_id=6364 > [2] > http://files.openinworld.com/Pharo-1.3-13307-OneClickWin-ODBC-thx-Mariano.z… > > cheers, Ben > > Mariano Martinez Peck wrote: > > > > On Tue, Dec 13, 2011 at 11:50 AM, Ben Coman <btc(a)openinworld.com> wrote: > >> Thanks, but nothing is attached. >> > > Sorry, here it is. > > >> Also, I'm not sure how to "make sure odbc libraries are findable by >> Windows." >> > > odbc.dll and friends should be in a place where Windows search dlls. I > think you won't have a problem since Windows usually puts such library in > its correct place. > > >> Nothing is changed in the environment between running 1.2.1 and 1.3. >> >> btw, I've been reading your blog on building the VM, and you indicated >> you would welcome feedback on english grammar. Here in one and a typo... >> http://marianopeck.wordpress.com/2011/04/ >> 1. "as you can image" >> "as you can imagine" >> 2. "wolrd" >> "world" >> >> > Thanks, I have just integrated them. > > >> cheers, Ben >> >> >> >> Mariano Martinez Peck wrote: >> >> Hi. Just as a test, would you like to test the attached VM? change xxx >> with zip. >> Just make sure odbc libraries are findable by Windows. >> >> Cheers >> >> On Sun, Dec 11, 2011 at 6:45 AM, Ben Coman <btc(a)openinworld.com> wrote: >> >>> Fadi Mansour wrote: >>> >>>> Hi, >>>> >>>> I'm having the same issue while trying to use ODBC with Pharo's OnClick >>>> image 1.3. >>>> >>>> Did you find anything about the issue here? >>>> >>>> Regards, >>>> >>>> Fadi >>>> >>>> >>>> >>>> >>>> >>>> >>> I've been offline while attending to a machine rebuild shutdown on a >>> mine site. I've only just got back today to start having a look at it >>> again. >>> >>> >> >> >> -- >> Mariano >> http://marianopeck.wordpress.com >> >> >> >> > > > -- > Mariano > http://marianopeck.wordpress.com > > > > -- Mariano http://marianopeck.wordpress.com

2 4

[Pharo-project] [update 1.4] #14259
by Marcus Denker Dec. 16, 2011

Dec. 16, 2011

14259 ----- Issue 5107: Warning: The method DialogItemsChooser>>openInWorld has been deprecated. http://code.google.com/p/pharo/issues/detail?id=5107 Issue 4949: Integer should be polymorphic with Fraction http://code.google.com/p/pharo/issues/detail?id=4949 -- Marcus Denker -- http://marcusdenker.de

1 0

Re: [Pharo-project] IdentitySet but using #hash rather than #identityHash ?
by Henrik Sperre Johansen Dec. 16, 2011

Dec. 16, 2011

On 16.12.2011 00:40, Henrik Sperre Johansen wrote: > On 15.12.2011 23:35, Mariano Martinez Peck wrote: >> Well...it is much slower :( it seems that the cost of (aKey >> identityHash + ( aKey mareaClass identityHash bitShift: 12) + (aKey >> basicSize bitShift: 24) >> is bigger than the colisions. >> Anyway, thanks for the nice thread. I learned. >> >> Cheers >> > Well, first of all, that always creates a largeInteger, since > identityHash is already shifted by 22... You'd want to be using > basicIdentityHash. > 2nd off, basicSize is useless for non-variable classes, which I guess > is the common case. > 3rd, putting class hash in the high bits won't really help much if > you're serializing many instances of the same class, as they will all > occupy a sequential hash range. > > So if I were you, I'd try with something like > obj basicIdentityHash << 12 + (obj class basicIdentityHash) > before discarding it outright due to computation cost. > > Cheers, > Henry > Note: This is not portable to Squeak. They kept the old version of identityHash (renamed to basicIdentityHash in Pharo), and introduced scaledIdentityHash (doing the same as identityHash in Pharo), due to backwards compatability concerns for users explicitly relying on #identityHash returning a value in range [0, 4095]. Cheers, Henry

2 1

Re: [Pharo-project] IdentitySet but using #hash rather than #identityHash ?
by Henrik Sperre Johansen Dec. 16, 2011

Dec. 16, 2011

On 16.12.2011 01:06, Henrik Sperre Johansen wrote: > > Usual identitySet: > n := 1000000. > set := IdentitySet new: n. > Time millisecondsToRun: [1 to: n do: [:waste | set add: Object new.]] > 4595 > > Custom subclass with a slightly modified hash (anObject identityHash + > (anObject class basicIdentityHash bitShift: 6)): > n := 1000000. > set := PseudoIdentitySet2 new: n. > Time millisecondsToRun: [1 to: n do: [:waste | set add: Object new.]] > 3694 > Why 6 you ask? Because I'm worth it. > > Not really enough to justify the extra compexity, imho. > > Cheers, > Henry > To put that a bit more in perspective, the extra cost for not preallocating to a sane value: n := 1000000. set := IdentitySet new. Time millisecondsToRun: [1 to: n do: [:waste | set add: Object new.]] 6872 Cheers, Henry

1 0

[Pharo-project] UnixFileDirectory default directoryExists: UnixFileDirectory default asString, '/.'
by Eliot Miranda Dec. 15, 2011

Dec. 15, 2011

Hi All, just noticed that UnixFileDirectory default directoryExists: UnixFileDirectory default asString, '/.' answers false, since directoryNames excludes '.' and '..' (justifiable, and changing this might break all manner of things). i.e. FileDirectory>directoryExists: filenameOrPath "Answer true if a directory of the given name exists. The given name may be either a full path name or a local directory within this directory." "FileDirectory default directoryExists: FileDirectory default pathName" | fName dir | DirectoryClass splitName: filenameOrPath to: [:filePath :name | fName := name. filePath isEmpty ifTrue: [dir := self] ifFalse: [dir := self directoryNamed: filePath]]. ^dir exists and: [ self isCaseSensitive ifTrue:[dir directoryNames includes: fName] ifFalse:[dir directoryNames anySatisfy: [:name| name sameAs: fName]]]. So how about FileDirectory class methods for accessing currentDirectoryNickname ^'.' FileDirectory>directoryExists: filenameOrPath "Answer true if a directory of the given name exists. The given name may be either a full path name or a local directory within this directory." "FileDirectory default directoryExists: FileDirectory default pathName" | fName dir | DirectoryClass splitName: filenameOrPath to: [:filePath :name | fName := name. filePath isEmpty ifTrue: [dir := self] ifFalse: [dir := self directoryNamed: filePath]]. ^dir exists and: [fName = self class currentDirectoryNickname or: [self isCaseSensitive ifTrue:[dir directoryNames includes: fName] ifFalse:[dir directoryNames anySatisfy: [:name| name sameAs: fName]]]] -- best, Eliot

1 1

[Pharo-project] Technical "Debt" of Smalltalk with Moose? (was Re: An Improvement Over its Successors)
by Janko MivÅ¡ek Dec. 15, 2011

Dec. 15, 2011

Hi guys, I'm wondering if we can produce the same comparison of Technical Debt with Moose technology comparing Smalltalk with Java, .NET and others, as in study below? >From study: Technical debt is highest in Java-EE, .NET second. Technical Debt represents the effort required to fix problems that remain in the code when an application is released. It is shown as cost per LOC. This year is by their study 3.61 USD in average. S, Janko MivÅ¡ek piÅ¡e: > S, Casey Ransberger piÅ¡e: > >> I read recently that a study done on enterprise apps written in many different languages found the most defects per capita in Java apps. This didn't come as a surprise, but what did: the apps with the lowest defect rates were written in COBOL. > > This study can be found here (registration needed): > > The CRASH Report - 2011/12 > (CAST Report on Application Software Health) > http://www.castsoftware.com/resources/resource/whitepapers/cast-report-on-a… > > Interesting read, recommended, findings can help improve our software too! > > >> The researchers seemed to scratch their heads and proclaim "...well, I guess there've been people grinding bugs off of the COBOL code for thirty years." >> >> After the horror stories that I read about on comp.lang.cobol (I have strange hobbies? Also, some of those COBOL guys have a lot to teach) I think I nearly fell out of my chair laughing and achieved something approaching enlightenment;) >> > -- Janko MivÅ¡ek Aida/Web Smalltalk Web Application Server http://www.aidaweb.si

1 0

Re: [Pharo-project] NB Bug?
by Fernando Olivero Dec. 15, 2011

Dec. 15, 2011

Your right, this isnt a bug, NativeBoost is very well tested and works really nice! I should reformulate the comment: 'if i new user of Native Boost, and i copy and paste some Library method, to create a NB method in smalltalk, and if it happens that that particular method is written foo(void), then will i be able to know the origin of the DNU ?, that i should just put foo() ?" For me it took a while of debugging before i understood the issue, maybe others less eager to debug will think its a bug? Surely with a good documentation and examples, it shouldn't be a problem. Fernando On Thu, Dec 15, 2011 at 1:23 AM, Igor Stasenko <siguctua(a)gmail.com> wrote: > On 15 December 2011 00:12, Fernando Olivero <fernando.olivero(a)usi.ch> wrote: >> I've insist. >> >> Because maybe somebody else runs into this bug, and thinks that NB is >> broken, instead of understanding they should use new C syntax. For >> examples i was using the latest Pango code. > > but this is NOT a bug. maybe a missing feature but not a bug. > Is it so hard to put empty braces instead of void? :) And besides > empty braces it is valid C syntax. > >> >> If you want we could raise a warning, suggesting changing the code? >> (maybe too disturbing). >> > no, i just dont understand why complicating a code for nothing? > You can define a callout to functions with no arguments. > If you would not be able to do so, then this can be considered a bug. > >> Fernando >> >> On Wed, Dec 14, 2011 at 2:10 AM, Igor Stasenko <siguctua(a)gmail.com> wrote: >>> On 14 December 2011 01:17, Fernando Olivero <fernando.olivero(a)usi.ch> wrote: >>>> Igor, is this a bug? >>>> >>>> When parsing methods that send native functions withouth arguments , >>>> void, theres a DNU. >>>> For example in the following method: >>>> >>>> ^ self pangoCairoNbCall:#(PangoCairoFontMap >>>> pango_cairo_font_map_get_default ( void ) ) >>>> >>>> >>>> I fixed it like this: >>>> >>>> usesMethodArguments >>>> "first check wether my single argument is void" >>>> ^(fnSpec arguments size = 1 ==> [fnSpec arguments first class ~= >>>> NBVoid]) and:[ fnSpec arguments anySatisfy: [:type | type loader >>>> usesSTStack ] ] >>>> >>>> >>> >>> Well, the function signature syntax are not strictly following C. >>> Just use empty parens for functions with no arguments. This is quite >>> archaic C syntax, which is really not important to support. >>> >>> Or i can add this patch if you insist. :) >>> >>> -- >>> Best regards, >>> Igor Stasenko. >> > > > > -- > Best regards, > Igor Stasenko.

3 2