July 2011 - Pharo-dev - lists.pharo.org

Re: [Pharo-project] image freezes when setting time "back"

by Mariano Martinez Peck

On Sat, Jul 16, 2011 at 8:07 PM, Philippe Marschall <kustos(a)gmx.net> wrote: > Hi > > I have had the following experience. I'm running Pharo 1.3 on a Linux > box where the time is set on start up using ntp. The hardware clock is > set to the local time (because I'm dual booting). When there is no > network connection during booting the system the clock is two hours > "ahead" until I get network connection and set the time using ntp. When > a Pharo image is running during this the UI freezes and doesn't react > anymore. Does anybody else observe this? Is this a known limitation? > > Yes. It seems the known limitation of Cog VM that uses clock and something related to heartbeat and etc...When you change time you confuse Cog generating a freeze. At least I remember that something like that could happens. Try with a non-Cog VM. Sorry if this message got sent twice. I had some trouble with the > mailing list. > > Cheers > Philippe > > > -- Mariano http://marianopeck.wordpress.com

July 16, 2011

[Pharo-project] image freezes when setting time "back"

by Philippe Marschall

Hi I have had the following experience. I'm running Pharo 1.3 on a Linux box where the time is set on start up using ntp. The hardware clock is set to the local time (because I'm dual booting). When there is no network connection during booting the system the clock is two hours "ahead" until I get network connection and set the time using ntp. When a Pharo image is running during this the UI freezes and doesn't react anymore. Does anybody else observe this? Is this a known limitation? Sorry if this message got sent twice. I had some trouble with the mailing list. Cheers Philippe

July 16, 2011

[Pharo-project] SerialPort>>openPort:

by Schwab,Wilhelm K

I am starting to build a 1.3 image, and taking the usual look around for things that would get lost. SerialPort>>openPort: appears to address named ports, but it looks like it will miss failures. What I did for 1.1 is capture the result of the primitives and raise an error if it is nil. Any interest? As an aside, does anyone know the state of serial ports on Linux? With Cog? I have been having good luck with native ports using Andreas' Windows VM. I have not done as well on my Linux laptop, but I can't tell whether that is related to Linux, the Linux VM, or (my lead theory) USB-serial converters. I need to get Ubuntu on a box with native ports and put it where I can stress test serial I/O; it has been a elusive combination, but I will get there. Bill

July 15, 2011

Re: [Pharo-project] A request or suggestion.

by StÃ©phane Ducasse

Check the settings because you can say to the window manager that it should stagger or position them in another way. Else have a look at TWM Stef > I'm still not realyl understanding the pharo Morph stuff. So I just > want to ask. Would it be possible/feasable to group windows. E.g I'd > have a group of Workspaces a group of Browsers etc. > > Currently if the windows just pop open it's hard to find a workspace one > may be interested. If there would be a group this should be much easier. > > Regards > Friedrich >

July 15, 2011

Re: [Pharo-project] Hanging connects, exhausted resources, memory leaks, bocked everything ... Unusable!

by Igor Stasenko

On 15 July 2011 12:39, Guido Stepken <gstepken(a)googlemail.com> wrote: > I am doing server tuning, building up fast server systems for a very long time now. > > Yesterday i installed 1.2.1, expecting to see a far more matured Pharo Smalltalk, stable, pretty nice new features. > > I opened up the Memory Grapher, hmmm.... works fine, watching memory pumping. A bit strange, seeing a Pharo consuming and freeing megabytes of memory space without Â any load. But ok! > > Then i opened up TestSuites, marked some tests, let them run. Ups! Desktop hanging. > > I expected to have with Pharo a well done, stable multitasking environment, but Pharo hangs everywhere, is leaking lots of memory, is keeping up connections, wasting resources unneccessarily, having problems with internal race conditions on semaphores, mutexes, feels same like "Big Kernel Lock", but even without MultiProcessing. > > The Pharo Team IMHO is very much missing important knowledge about modern Operating System Design, what Pharo Smalltalk OS is in fact, without being able to fulfill the minimum standards i expect from a usable Smalltalk OS. > > As system operator or enterprise i cannot offer such a system, that has to be restarted twenty times a day, due to hangs, memory leaks, sucking connection handles, ..., whatever! > > tnx for your understanding! > > regards, Guido Stepken > >From an outsider point of view, your message is ok: "i took a glance on system and found it inappropriate for my tasks, so i will not use it". Fine. Now, from an insider point of view, your message is worthless, because it doesn't contains any concrete proposals to make things better (to your thinking). We would like to improve our system to make it good enough for people like you. But we need help. So, every time you saying 'i don't like X', you should also argument that and propose better solution. If you not doing that, but instead using vague statements, you just wasting own and our time. Lets be concrete and constructive. Pharo is far from perfect. And everyone understands that. You made no discovery. Now, ask yourself how your message helps to improve that? -- Best regards, Igor Stasenko AKA sig.

July 15, 2011

[Pharo-project] A request or suggestion.

by Friedrich Dominicus

I'm still not realyl understanding the pharo Morph stuff. So I just want to ask. Would it be possible/feasable to group windows. E.g I'd have a group of Workspaces a group of Browsers etc. Currently if the windows just pop open it's hard to find a workspace one may be interested. If there would be a group this should be much easier. Regards Friedrich

July 15, 2011

Re: [Pharo-project] Hanging connects, exhausted resources, memory leaks, bocked everything ... Unusable!

by Sven Van Caekenberghe

On 15 Jul 2011, at 13:50, Janko MivÅ¡ek wrote: > Before shooting the messenger of bad news let we rather acknowledge that > there are problems in Squeak/Pharo on the socket/semaphore/process > field. I also admit, I somewhat lost nervers on Pharo as Guido did and > I'd like to post the same post. But of course bringing bad messages must > always be done diplomatically of you won't like to be shoot :) Janko, you have an excellent Smalltalk reputation and your comments, feedback and contribution are positive. You walk the talk. We take you seriously. > So, known problems are: socket limit over about 200, semaphores (i > remember discussion back, is this solved?) protecting critical sections > doesn't work well, processes? After our last discussion, I added ! "Extend the external object table to allow for more sockets" Smalltalk vmParameterAt: 49 put: 8192. ! to my server startup scripts. This should be good for at least around 2000 sockets simulataneously, if I understand things correctly. (Your OS has to allow that many for a single process as well ;-) Is there a bug report/issue and/or testcase for the critical section problem ? Regards, Sven

July 15, 2011

Re: [Pharo-project] Fixed up AutomaticMethodCategorizer

by StÃ©phane Ducasse

BTW I would love to recategorize all the methods in the image :) Stef On Jul 15, 2011, at 1:48 PM, Mariano Martinez Peck wrote: > Thanks Danie. I don't have time to take a look right now, but maybe someone can. Anyway, you may know AutomaticMethodCategorizer much better than us. So you can just go ahead and commit and update ConfifugrationOfAutomaticMethodCategorizer > > On Wed, Jul 13, 2011 at 5:27 PM, Danie Roux <danie(a)danieroux.com> wrote: > Hi Mariano, > > We need the capability to set custom rules on > AutomaticMethodCategorizer. I've added it back in > http://www.squeaksource.com/autoMethodCat/AutomaticMethodCategorizer-DanieR… > > If you will please test it and update the Configuration, I will appreciate it. > > -- > Danie Roux *shuffle* Adore Unix - http://danieroux.com > Email / Google Talk / Jabber - danie(a)danieroux.com > > > > -- > Mariano > http://marianopeck.wordpress.com >

July 15, 2011

Re: [Pharo-project] Hanging connects, exhausted resources, memory leaks, bocked everything ... Unusable!

by Norbert Hartl

Am 15.07.2011 um 12:39 schrieb Guido Stepken: > As system operator or enterprise i cannot offer such a system, that has to be restarted twenty times a day, due to hangs, memory leaks, sucking connection handles, ..., whatever! Sorry, but that made me laugh. And I'm really asking myself if you ever worked in big industries. If you think like this you shouldn't ever dare to, you will asking for nightmares. couldn't resist, Norbert

July 15, 2011

Re: [Pharo-project] Hanging connects, exhausted resources, memory leaks, bocked everything ... Unusable!

by Janko MivÅ¡ek

Hi all, Before shooting the messenger of bad news let we rather acknowledge that there are problems in Squeak/Pharo on the socket/semaphore/process field. I also admit, I somewhat lost nervers on Pharo as Guido did and I'd like to post the same post. But of course bringing bad messages must always be done diplomatically of you won't like to be shoot :) So, known problems are: socket limit over about 200, semaphores (i remember discussion back, is this solved?) protecting critical sections doesn't work well, processes? So Guido is right, don't shut them, let his anger rather energitize us to finally solve them! Janko S, StÃ©phane Ducasse piÅ¡e: > Sven from beta9 is using Pharo as a server and it works well from what he told us during the last sprint. > Netstyle is also using pharo for Seaside. > > Now there is certainly place for improvements and pharo is improving every day (but because there are cool pharoers improving it). > > I noticed in the past that you always have this same aggressive tone. > And even worse, we value critics and non politically correct discussions but not systematic bashing. > What I dislike is the tone of your email. So do not continue like that if you want to have any support for people. > else you will be plonkified :) > >> I am doing server tuning, building up fast server systems for a very long time now. >> >> Yesterday i installed 1.2.1, expecting to see a far more matured Pharo Smalltalk, stable, pretty nice new features. >> >> I opened up the Memory Grapher, hmmm.... works fine, watching memory pumping. A bit strange, seeing a Pharo consuming and freeing megabytes of memory space without any load. But ok! >> >> Then i opened up TestSuites, marked some tests, let them run. Ups! Desktop hanging. >> >> I expected to have with Pharo a well done, stable multitasking environment, but Pharo hangs everywhere, is leaking lots of memory, is keeping up connections, wasting resources unneccessarily, having problems with internal race conditions on semaphores, mutexes, feels same like "Big Kernel Lock", but even without MultiProcessing. > > Only that. Not bad. We are soo dumb that it could be so worse. > >> The Pharo Team IMHO is very much missing important knowledge about modern Operating System Design, what Pharo Smalltalk OS is in fact, without being able to fulfill the minimum standards i expect from a usable Smalltalk OS. >> >> As system operator or enterprise i cannot offer such a system, that has to be restarted twenty times a day, due to hangs, memory leaks, sucking connection handles, ..., whatever! >> >> tnx for your understanding! > > My understanding is that you do not know how to communicate. > > Now my advice is to try 1.3 because lot of improvements were applied and for example Zinc is a good starting part. > >> >> regards, Guido Stepken > > > -- Janko MivÅ¡ek Aida/Web Smalltalk Web Application Server http://www.aidaweb.si

July 15, 2011