Squeak-dev April 2009

squeak-dev@lists.squeakfoundation.org

104 participants
168 discussions

Q: Contributing to Squeak
by K. K. Subramaniam 20 May '09

20 May '09

Hi, If one wants to contribute to Squeak but uses Etoys/Linux most of the time, where should the contributions be made? - etoys in dev.laptop.org - squeak mainline (bugs.squeak.org) - etoys in tracker.squeakland.org Will Etoys ever merge into mainline Squeak (say in Squeak 4.0). Will it then appear as a changeset that can be applied to Squeak rather than a separate image? TIA .. Subbu

8 12

Status of translations before Squeak 4.0
by Séverin Lemaignan 19 May '09

19 May '09

Hello everyone, I was wondering what was the status of the translation workflow in the upcoming versions of Squeak? to be more precise, I would like to update the (far from perfect) French translation from what I did for the Etoys image, and I'm wondering what is the "procedure". Are we heading to a "100% gettext" approach like the OLPC one (which seems very efficient to me)? Cheers, Severin

4 5

Starting Smalltalk with Squeak problem with color selector
by Tom18 19 May '09

19 May '09

Hello all, I am starting programming in smalltalk and I would like to create a morph in which I would like to insert a color selector that changes the color of my morph. Maybe it is easy to do but I have tried many times without success. Thanks for the future responses. -- View this message in context: http://www.nabble.com/Starting-Smalltalk-with-Squeak-problem-with-color-sel… Sent from the Squeak - Dev mailing list archive at Nabble.com.

4 5

Problem with cdelc and apicall
by Mariano Martinez Peck 13 May '09

13 May '09

People: I am having a problem with cdelc and apicall in windows with SqueakDBX. I have a abstract class OpenDBX with the subclasses: OpenDBXWin32 and OpenDBXUnix. SqueakDBX detects which OS you are using and instances one of them. They have these methods: OpenDBXWin32: apiBind: handle database: databaseName name: userName password: password method: method "int odbx_bind(odbx_t* handle, const char* database, const char* who, const char* cred,int method )" <apicall: long 'odbx_bind' (ulong char* char* char* ulong) module: 'libopendbx-1.dll'> ^self externalCallFailed OpenDBXUnix: apiBind: handle database: databaseName name: userName password: password method: method "int odbx_bind(odbx_t* handle, const char* database, const char* who, const char* cred,int method )" In linux, I don't have problems, but in windows, when I try to download it from MC, I have a "sintax errror" "OpenDBXUnix api calls apiBind: handle database: databaseName name: userName password: password method: method" apiBind: handle database: databaseName name: userName password: password method: method "int odbx_bind(odbx_t* handle, const char* database, const char* who, const char* cred,int method )" <> expected ->cdecl: long 'odbx_bind' (ulong char* char* char* ulong) module: 'opendbx'> ^self externalCallFailed <cdecl: long 'odbx_bind' (ulong char* char* char* ulong) module: 'opendbx'> ^self externalCallFailed Does anyone have an idea of what can be happening ? thanks in advance. Mariano

2 8

Contributing to 3.11: Step-by-step
by Andreas Raab 11 May '09

11 May '09

Folks - To summarize some of the discussion on IRC, here is a step-by-step guide for how to contribute to 3.11: 1. Report an issue at http://bugs.squeak.org/ 2. If you have a fix/patch attach it to the bug report 3. Add an installer script to the notes, like here: (for more options see http://installer.pbwiki.com/Installer#Mantis) "fix begin" Installer mantis bug: 1234 fix: 'FixTheBug.cs'. "fix end" 4. Test your fix: Installer mantis ensureFix: 1234 5. To propose this fix as a candidate for 3.11: a) Obtain "developer" status on Mantis. You can do this by mailing box-admins(a)lists.squeakfoundation.org. b) Update the issue and set the "Status" field to "testing" c) Update the issue and set "Fixed in Version" field to "3.11" This field is only accessible in the "advanced" view, so if you do not see it you need to change to the "advanced" view by clicking on "view advanced" or "update advanced" in the top right. 6. Load all the fixes with testing status: (Installer mantis bugsAll select:[:bug| ((bug status = 'testing' or:[bug status = 'resolved']) and:[bug fixedIn = '3.11']]) do:[:bug| bug ensureFix]. 7. Rinse and repeat. Cheers, - Andreas

3 4

Suspending process fix
by Igor Stasenko 08 May '09

08 May '09

Hello, i trying to provide a fix to the issue http://bugs.squeak.org/view.php?id=6822 what is needed to make following code work w/o error: |sema proc | sema := Semaphore new. proc := [ sema wait. self error: 'should not be there' ] fork. Processor yield. proc suspend. proc resume. The problem here, is that once you issue a #suspend on a process it puts nil to its myList ivar, ignoring the fact that it could be waiting on semaphore. This leads to disconnecting the process from semaphore, and any subsequent 'sema signal' will be simply lost: suspend | oldList | <primitive: 88> "primitive will fail, if receiver is not active process, which is just our case" myList ifNil:[^nil]. oldList := myList. myList := nil. oldList remove: self ifAbsent:[]. ^oldList the possible fix would be: - add an instance variable to Process, say 'semaphore'. - fix the #suspend to following: suspend | oldList | <primitive: 88> "primitive will fail, if receiver is not active process, which is just our case" myList ifNil:[^nil]. oldList := myList. myList := nil. oldList remove: self ifAbsent:[]. oldList isKindOf: Semaphore ifTrue: [ semaphore := oldList ]. ^oldList - then, in the #resume resume suspendedContext ifNil: [^ self primitiveFailed]. semaphore ifNotNil: [ semaphore resumeWaiting: self. semaphore := nil. ^ self. ] ^ self primitiveResume and Semaphore>>resumeWaiting: aProcess excessSignals>0 ifTrue: [excessSignals := excessSignals-1. aProcess primitiveResume. "resume immediately" ] ifFalse: [self addLastLink: aProcess] ---------- An attached file is the changeset which makes following test to work w/o failure: |sema proc bool | bool := true. sema := Semaphore new. proc := [ sema wait. bool ifTrue: [self error: 'should not be there']. Transcript show: 'semaphore signaled' ] fork. Processor yield. proc newSuspend. proc newResume. self assert: (sema last == proc). bool := false. sema signal. P.S. i didn't checked how this fix will affect other things around processes, like mentioned by Mike: Note that Process>>signalException: (as of in Squeak 3.8, it may be fixed in later versions?) relies on this broken behaviour. -- Best regards, Igor Stasenko AKA sig.

7 33

Socket clock rollover issues
by Andreas Raab 06 May '09

06 May '09

Hi - I'm not sure anyone cares about this (most likely Seaside users), but class Socket has some issues with clock-rollover which can be seen on high load servers that have enough uptime and load. The problem is caused by Socket>>deadlineSecs: which at the time of the clock rollover computes a deadline that can never be reached. If you've seen connections that wouldn't time out apparently at random there is a chance you have been affected by this type of issue. Bug (and fix) are posted at http://bugs.squeak.org/view.php?id=7343 Cheers, - Andreas

3 7

[ANN] Aida/Web 6.0 second beta is out!
by Janko Mivšek 04 May '09

04 May '09

Dear Smalltalkers, That time with a considerable more fresh meat and on both Squeak+friends and VisualWorks, a second beta of Aida/Web 6.0 web framework and application server is released. To download it visit our renewed website http://www.aidaweb.si With that beta we also changed the direction: now the 6.0 won't be just a companion of Scribo CMS but a standalone release with support of tree-like control flow, so that this release will really deserve a major number jump from 5.6 to 6.0. Here are just 3 most important new features in that beta: 1.Introduction of web execution contexts as a step towards tree-like control flow. These contexts give a scope to the "execution" of web request and are the precondition for controlling this flow of execution. 2.Translation support for text on the web apps, to allow the multilingual text on the pages, besides multilingual content, 3.Caching (memcached like) to improve performance of serving the content, which otherwise doesn't change frequently. But much more in release notes: http://www.aidaweb.si/release-notes-6.0.html Best regards Janko Mivšek -- Janko Mivšek AIDA/Web Smalltalk Web Application Server http://www.aidaweb.si

2 2

FFI question [WAS] squeakopendbx, which VM?
by Mariano Martinez Peck 04 May '09

04 May '09

Hi folks! I have a problem with SqueakDBX and FFI. I install openDBX from a .deb and I get this: ldconfig -p | grep opendbx show the same as you: libopendbxplus.so.1 (libc6) => /usr/lib/libopendbxplus.so.1 libopendbx.so.1 (libc6) => /usr/lib/libopendbx.so.1 The problem is that we (in squeakDBX) have as module: opendbx. Linux looks for libMODULE_NAME.so. In this case, was searching libopendbx.so. But, my file was named libopendbx.so.1. And of course, I got a 'unable to find function address'. Solution: copy and rename (or just rename) libopendbx.so.1 to libopendbx.so and worked like a charm. But, the question is, shouldn't FFI resolve this ? is this a matter of the linker ? Thanks for the help! Mariano ---------- Forwarded message ---------- From: Mariano Martinez Peck <marianopeck(a)gmail.com> Date: Tue, Apr 28, 2009 at 10:40 PM Subject: Re: [Pharo-project] squeakopendbx, which VM? To: Pharo-project(a)lists.gforge.inria.fr I think I fix it :) I took a clean Ubuntu. I did: sudo dpkg -i libopendbx1_1.4.0-1_i386.deb libopendbx1-mysql_1.4.0-1_i386.deb Then ldconfig -p | grep opendbx show the same as you: libopendbxplus.so.1 (libc6) => /usr/lib/libopendbxplus.so.1 libopendbx.so.1 (libc6) => /usr/lib/libopendbx.so.1 The problem is that we have as module: opendbx. Ubuntu looks for libMODULE_NAME.so. In this case, was searching libopendbx.so. But, your (and my) file was named libopendbx.so.1. Solution: copy and rename (or just rename) libopendbx.so.1 to libopendbx.so and should work like a charm. Thanks for your testing Hilaire. We want SqueakDBX to be as much easier as possible. So, documenting the wiki with several cases will definitively help us. I add this information to the wiki. Let me know if that work. Cheers, Mariano On Tue, Apr 28, 2009 at 2:56 PM, Mariano Martinez Peck < marianopeck(a)gmail.com> wrote: > > > 2009/4/28 Hilaire Fernandes <hilaire(a)ofset.org> > >> >> >> 2009/4/28 Mariano Martinez Peck <marianopeck(a)gmail.com> >> >>> >>> >>>> I don't have it in my version of Pharo. I don't know if I can update it >>>> without going through a full pharo image which I don't want to do now. >>>> >>> >>> >>> Ok. Doesn't matter. Your current problem is not about FFI but you will >>> have that problem then (when you get squeakDBX working). You only need to >>> get the latests FFI version from MC to work in pharo with closure VM. >>> >> >> I am pretty sure the VM I used is not with the fixed closure. I am note >> sure about the image, I can tell you tomorrow the version number of the >> image when at the office. >> Is the fixed closure mandatory to get SqueakDBX working? >> Which Pharo # image and VM should be used? >> > > SqueakDBX works with both. If you use a closure image you must use a VM > with closure an latest FFI (from MC). But, you can also use Squeak in older > version (non closure) with FFI from Universes. So, no problem for you. > > > >> >> I haven't my linux machine here, I will look tonight in my home, but: >> >> > 1) what do you have /usr/lib in PATH ? >> No it points only to binary as it should. >> Does FFI looks at PATH for the libraries? If so it is a buggy behavior. >> I am really currious to know how FFI find the libraries... >> > > I am not sure about this. Perhaps someone (Andreas) can help us. > > >> >> >> 2) Just for testing, can you put the openDBX libraries in the same >> directory where the pharo >> image is. >> >> 3) The same as 2) but where the VM is. >> >> I will try tomorrow morning. >> > > Which Linux are you using? Perhaps there is a problem trying to resolve > the library. The last test you can do (just to see if it works) is this: > > Go t class OpenDBXUnix and change the methods > > apiBind: handle database: databaseName name: userName password: password > method: method > > and > > apiInitialize: handle backend: backend host: host port: port > > In both, put module: 'libopendbx.so.1' or module: 'libopendbx.so' instead > of module: 'opendbx' > > I hope you get it working. Ask all what you want. > > Greetings, > > Mariano > > >> >> >> Thanks! >> >> >> Hilaire >> >> >> -- >> http://blog.ofset.org/hilaire >> >> _______________________________________________ >> Pharo-project mailing list >> Pharo-project(a)lists.gforge.inria.fr >> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project >> > >

3 10

Some gotchas with #terminate and #critical:
by Igor Stasenko 02 May '09

02 May '09

Currently i replacing all calls which affect the scheduling to redirecting them to Processor, i.e. instead: self primitiveDoSomething do: Processor doSomethingWith: self. This is to ensure, that no code, in rest of classes , except ProcessorScheduler deals with scheduling directly. If i change the Semaphore>>wait method in that way, then it breaks the trick which sitting in #terminate method : "Figure out if we are terminating the process while waiting in Semaphore>>critical: In this case, pop the suspendedContext so that we leave the ensure: block inside Semaphore>>critical: without signaling the semaphore." (inSema == true and:[ suspendedContext method == (Semaphore compiledMethodAt: #critical:)]) ifTrue:[ suspendedContext := suspendedContext home. ]. What you think is following alternative implementation: critical: mutuallyExcludedBlock | blockValue caught | caught := false. [ caught := true. [self wait.] ifCurtailed: [ caught:=false ]. blockValue := mutuallyExcludedBlock value ] ensure: [caught ifTrue: [self signal]]. ^blockValue avoids the need in having tricks in #terminate? -- Best regards, Igor Stasenko AKA sig.

2 4

Jump to page: