Squeak-dev May 2009

squeak-dev@lists.squeakfoundation.org

102 participants
131 discussions

Squeak Oversight Board Meeting Report for 5/7/2009
by Ken Causey 08 May '09

08 May '09

Please read the report at http://squeakboard.wordpress.com/2009/05/08/meeting-report-for-572009/ and let us know what you think. Ken

1 0

RE: [squeak-dev] Cuis on Windows Mobile (PDA)
by Giuseppe Luigi Punzi Ruiz 08 May '09

08 May '09

Could be good. My intention, is to use the browser to model when out of the office, and the menu (do it, accept, inspecting etc?) with tap pressed could be very usefull. Other issue could be usefull is something for keyboard,but this is other history. Enviado usando Real Mail de Vodafone. -----Original Message----- From: "Juan Vuletich" <juan(a)jvuletich.org> Sent: Fri, 08 May 2009 09:13:52 -0300 To: "The general-purpose Squeak developers list" <squeak-dev(a)lists.squeakfoundation.org> Received: 08-May-2009 14:14:18 +0200 Subject: Re: [squeak-dev] Cuis on Windows Mobile (PDA) Hi Giuseppe! Giuseppe Luigi Punzi wrote: > Hi, > > I'm trying Cuis on my HTC, and works with very very good performance. Good! > A little problem I see, is, thath a Browser, when "maximized" don't > maximize to all world, there are always a margin, and, this margin, > for a screen in 320x240, is very big. How can I change this behaviour > and, when maximize, do it to all world withou any margin? Just set preference #fullScreenLeavesDeskMargins to false. Besides, if you use the screen rotated, and fonts look horrible, you might set #subPixelRenderFonts to false too. You might also tweak Preferences class >> scrollbarThickness to make them narrower. And I guess you already selected smaller fonts... > Other improvement, could be to launch the menu, when click, and stay > pressed (the default "right click" on Windows Mobile). Could be > possible to implement this? > > Cheers. Good idea! I'm working on it. I'll publish it during this weekend. Cheers, Juan Vuletich

1 0

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

macintosh carbon vm 4.1.1b2
by John M McIntosh 07 May '09

07 May '09

I've pushed out a macintosh carbon closure VM 4.1.1b2 for testing. It can be found via links on http://www.smalltalkconsulting.com/squeak.html It includes the following fixes. 4.1.1b2 Window Title now accepts UTF8 strings Build with VMMaker-dtl-121 includes fix for closure vm crash http://bugs.squeak.org/view.php?id=7349 primitiveDisableFileAccess primitiveDisableImageWrite primitiveDisableSocketAccess primitiveHasSocketAccess includes fix for Reference Mantis 6987: signed32BitValueOf:, signed64BitValueOf: etc. broken a fix for #signed32BitIntegerFor: on 64-bit images. includes fix fortwo methods with assignment to block temp arguments, reported on vm-dev list <http://lists.squeakfoundation.org/pipermail/vm-dev/2009-April/002607.html > 4.1.1b2 distributed only for testing to a few people. 4.1.0b1 UTF-8 on clipboard primitives via code from qwaq UTF-8 for file names compile SDK for ppc is 10.4.x now, versus 10.3.x Unknown if clipboard works on 10.2 or 10.3 ppc systems? -- = = = ======================================================================== John M. McIntosh <johnmci(a)smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com = = = ========================================================================

1 0

Odd performance characteristics of symbol creation
by Adrian Lienhard 07 May '09

07 May '09

Hi, We have a server application that, after some longer usage, becomes very slow. It turned out that this is because lots of symbols are created from strings (without holding onto them), which becomes slower and slower over time. This is easy to fix in our app, but I'd like to understand why Squeak has a problem with that. Using the following snipped, I produced the attached chart (Y axis is ms spent to create 5000 random symbols). You can nicely see the reoccurring pattern that grows over time. 1000 timesRepeat: [ Transcript print: ([ 5000 timesRepeat: [ (($a to: $x) atRandom asString , 'asdasd') asSymbol ] ] timeToRun); cr; flush. ] Message tally sais: 96.2% {13872ms} ByteString(String)>>asSymbol |96.2% {13872ms} Symbol class>>intern: | 49.7% {7167ms} Symbol class>>lookup: | |49.7% {7167ms} WeakSet>>like: | | 49.7% {7167ms} WeakSet>>scanFor: | | 32.3% {4658ms} primitives | | 17.3% {2495ms} UndefinedObject(Object)>>= | 44.4% {6402ms} WeakSet>>add: | |44.3% {6388ms} WeakSet(Set)>>findElementOrNil: | | 44.3% {6388ms} WeakSet>>scanFor: | | 22.8% {3288ms} UndefinedObject(Object)>>= | | 21.5% {3100ms} primitives | 2.1% {303ms} ByteSymbol>>string: 2.9% {418ms} Character>>to: 2.4% {346ms} SmallInteger(Number)>>to: 2.4% {346ms} Interval class>>from:to:by: 2.4% {346ms} Interval>>setFrom:to:by: I haven't had time to look into this very closely yet but maybe somebody else already knows what is going on. My guess is that the NewSymbols weak set is repeatedly grown since tally is not decreased if the GC removes a weak element. This is a Squeak 3.9 image (OS X and Linux). Cheers, Adrian ___________________ http://www.adrian-lienhard.ch/

5 15

Re: [squeak-dev] Problem with SMTPClient
by "S.J.Chun" 07 May '09

07 May '09

Hi, I hope this can help you. Can you let us know the log of the SMTP server in 127.0.0.1(that's what you tested with linux console, right?) ? Following code is one of my method for sending email messages, and I also cannot find any meaningful differences between your code and mine. sendMail | client msg m sub mm recvs | client _ SMTPClient openOnHostNamed: smtpServer. userID isEmptyOrNil ifFalse: [ client user: userID. password isEmptyOrNil ifTrue: [ client password: '' ] ifFalse: [ client password: password ]]. m _ MimeConverter forEncoding: 'base64'. "sub _ (m mimeEncode: (subject convertToWithConverter: (UTF8TextConverter new)) readStream) contents." sub _ (m mimeEncode: subject readStream) contents. sub _ sub copyReplaceAll: String cr with: ''. "mm _ (m mimeEncode: (messageBody convertToWithConverter: (UTF8TextConverter new)) readStream) contents." mm _ (m mimeEncode: messageBody readStream) contents. recvs _ (receivers inject: '' into: [ :s :e | s size = 0 ifTrue: [ s _ s, e ] ifFalse: [ s _ s, ', ', e ]]). msg _ 'MIME-Version: 1.0', String cr, 'Content-Type: text/html; charset=UTF-8', String cr, 'Content-Transfer-Encoding: base64', String cr, 'X-Priority: 3', String cr, 'X-Mailer: NXT', String cr, 'Sender: ', replyTo, String cr, 'Subject: ', '=?UTF-8?B?', sub, '?=', String cr, 'To: ', recvs, String cr, String cr, mm. [ client login. client mailFrom: replyTo to: receivers asArray "text: (msg convertToWithConverter: (UTF8TextConverter new))." text: msg. client quit ] ensure: [ client close ]. ----- Original Message ----- From: Mariano Martinez Peck <marianopeck(a)gmail.com> To: The general-purpose Squeak developers list <squeak-dev(a)lists.squeakfoundation.org> Sent: 09-05-06 22:26:37 Subject: [squeak-dev] Problem with SMTPClient Hi: I am trying to send an email from Squeak. I have ubuntu and postfix smtp server running in my local machine. I know postfix is running OK. I could do successfully the following test from a linux console: 220 Servidor ESMTP HELO localhost 250 Hello, please meet you MAIL FROM: <a href="mailto:destinomochila@gmail.com" target="_blank">destinomochila(a)gmail.com</a> 250 Ok RCPT TO: <a href="mailto:matias_dolian@hotmail.com" target="_blank">matias_dolian(a)hotmail.com</a> 250 Ok DATA 354 End data with <CR><LF>.<CR><LF> Subject: Campo de asunto From: <a href="mailto:destinomochila@gmail.com" target="_blank">destinomochila(a)gmail.com</a> To: <a href="mailto:matias_dolian@hotmail.com" target="_blank">matias_dolian(a)hotmail.com</a> Hola, Esto es una prueba3. Adi?. . quit However, when I try to do SMTPClient I do something like this: | message sender recipients | message := 'From: DestinoMochila <<a href="mailto:destinomochila@gmail.com" target="_blank">destinomochila(a)gmail.com</a>> To: You <<a href="mailto:marianopeck@gmail.com" target="_blank">marianopeck(a)gmail.com</a>> Subject: Simple Mail from Squeak This is a test mail'. sender := '<a href="mailto:destinomochila@gmail.com" target="_blank">destinomochila(a)gmail.com</a>'. recipients := #('<a href="mailto:marianopeck@gmail.com" target="_blank">marianopeck(a)gmail.com</a>'). SMTPClient ??? deliverMailFrom: sender ??? to: recipients ??? text: message ??? usingServer: '127.0.0.1' I have no errors, but the email is never send :( This is the first time I try to use SMTPClient so perhaps I am doing something wrong. Any idea? Thanks in advance, Mariano

1 0

Re: [squeak-dev] Problem with SMTPClient
by "S.J.Chun" 07 May '09

07 May '09

1 0

ReferenceStream Problem
by Zulq Alam 06 May '09

06 May '09

When deserializing Dictionary objects, it looks like they are being rehashed. Where and why is this happening? dict1 := Dictionary new. 1 to: 1000 do: [:e | dict1 at: e put: e]. rep1 := ReferenceStream streamedRepresentationOf: dict1. dict2 := ReferenceStream unStream: rep1. rep2 := ReferenceStream streamedRepresentationOf: dict2. dict1 = dict2. " true " rep1 = rep2. " false " dict1 array size. " 2560 " dict2 array size. " 1334 " dict1 rehash. dict1 array size. " 1334 " I'm using ReferenceStreams to store object graphs in a relational database as a BLOB using Glorp. Glorp determines whether to do an updated by comparing the serialized form of objects with their last state. This causes spurious UPDATEs. Any ideas? Thanks. Zulq.

2 1

Very thanks !!!!
by Edgar J. De Cleene 06 May '09

06 May '09

People: Lucky us, we have dedicated members patient with alien old morons like me. Very thanks Ken, I could not only connect via Terminal , also with Cyberduck for put files on server. Now I waiting Andreas, I send a private mail to he some time ago. What I should do for help with 3.11, 4.0 or another release image ? I chatting with Juan Vuletich for learn from he and his Cuis. Maybe is time to all work on same thing... Edgar

1 0

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

← Newer
1
...
8
9
10
11
12
13
14
Older →

Jump to page: