Squeak-dev July 2007

squeak-dev@lists.squeakfoundation.org

140 participants
214 discussions

[ANN][Squeak-dev Beta Image] Version sq3.10-7130dev07.07.1
by Damien Cassou 01 Aug '07

01 Aug '07

Squeak-dev image is an image made for squeak developers. *Important notice:* This version is based on Squeak 3.10 7130. You can download it here: http://damien.cassou.free.fr/squeak-dev/beta/ This version contains a lot of new and exciting features: - OmniBrowser (OB) finally selects the --All-- method category per default - icons and colors are now possible on each node - the mercury panel (input field on top of all OB browsers) is now really usable: there is a balloon with documentation, the browser which appears is a specific browser with lots of cool features, dedicated to searches. For example, you can for OB*Browser to open a search browser on all classes whose names match OB*Browser. - the sender and implementor browsers display messages that are not sender or implementor anymore "struck out" - "Find method wildcard" is now available in OB too - A context menu on the packages now let you select amongst recently browsed classes - OB is more portable (a GemStone port is being developed) - Under the hood massive changes - Lots of bug fixes and code cleaning - eCompletion does not consider < to be a smart character anymore I would like to thank the following people for their recent and impressive work on the OmniBrowser framework (alphabetically): Cédrick Béler Colin Putney Duncan Mak David Mitchell David Röthlisberger Piers Cawley (I hope I do not forget anyone) More features will soon be available for OmniBrowser : - No more package categories on the first pane: now, it will contain Monticello packages and you will be able to publish, load, merge directly from OB. - UML diagrams automatically generated for your code - ... Contained packages: ------------------------ This image contains the following packages: AST version 144 AutomaticMethodCategorizer version 0.24 AutomaticMethodCategorizerServices version 0.2 ClassSelectorSets version 1 DynamicProtocols version 0.53 eCompletion version 0.88 eCompletion-Traits version 0.1 eCompletionOmniBrowser version 0.2 Installer version 1.0.92 OmniBrowser version 0.380 OmniBrowser-Full version 0.24 OmniBrowser-Morphic version 0.23 OmniBrowser-Refactory version 0.27 OmniBrowser-Standard version 0.271 OmniBrowser-SUnitIntegration version 0.2 OmniBrowser-TraitsIntegration version 0.1 Refactoring Engine version 37 RoelTyper version 0.60 ScriptManager version 0.6 Shout version 3.15-tween.70 ShoutWorkspace version 1-tween.1 SmaCC runtime version 13 Squeak dev packages version 0.13 SUnit version 0.33 Universes version 32 Universes OmniBrowser version 0.32 YAXO version 9 pdating the image: ------------------------- You are not obliged to download a new image if you don't want. You can always update your image. To upgrade your already existing squeak-dev image: 1) World menu->open...->Package Universe Browser 2) Update list from network 3) Select all upgrades 4) Install selection 5) Execute: 'DEVImageCreator default install' (all windows will be closed) Prepared packages: -------------------------- There are a lot of packages prepared for you in this image. They are not loaded in the image, but they can be easily. - Open Package Universe Browser (world->open...->Package Universe Browser). - Click on "Update list from network" - Wait a few seconds - Select your packages - Install them -- Damien Cassou

7 15

Squeak Swiki password issues.
by Blake 31 Jul '07

31 Jul '07

Hey, Swiki-knowledgeable types: I set up a swiki for someone who has locked a page and doesn't remember the password. I don't recall changing the default admin name password but "admin" and "password" don't work, nor do any of the likely combinations I would have used. Now what? I presume I can issue a command from inside the running image (which I do have access to). I can't find anything useful on the web about it, though. ===Blake===

2 3

Image freezing problem
by Adrian Lienhard 31 Jul '07

31 Jul '07

Hi, We recently run into the image freezing problem, where connecting with VNC and moving the mouse would bring the image alive again. When being stuck, Squeak hoggs the CPU, memory consumtion is stable. There have been various reports matching the described symptoms, e.g., http://lists.squeakfoundation.org/pipermail/squeak-dev/2006- August/107618.html. When seeing Andreas' delay issue, it looked just obvious that it's exactly this same problem. But the fix does not resolve it. So, after quite a lot of hours of debugging, here is an explanation of what is going on (kudos to Marcus, who helped me with putting the last pieces of the puzzle together). The problem occurred when producing a PDF of around 800 pages in a background process (priority 25). In production, this process runs besides serving a Seaside application (running with priority 30) and involves accessing the DB (PostgresQL). However, I managed to come up with a reproducable setup without involving Seaside requests and without accessing the DB. All there is: a CPU and memory intensive (~130MB) background process which does some logging to the file system. Setup: ----------------------------------------- Problem appeared in production with an older VM, so I decided to build a new one with the most recent fixes: - VM built from VMM 3.8b6 from SM + VmUpdates-100[1-6]-dtl.1.cs + JMM- VmUpdates32bitclean.2.cs - Squeak 3.7 image + Andreas' delay fix - gcc 2.95.4, Linux 2.4.18-686-smp #1 SMP i686 - Squeak is started headfull in VNC For tracing, I added the TraceMessageSends instrumentation from John and extended it with some more data. (BTW, I vote to integrate this by default as it proved, with the signaling facility, very useful!) Explanation for the cause: ----------------------------------------- The following message trace was started when the image got stuck and it was stopped again a few seconds after Squeak continued working normal again: http://www.netstyle.ch/dropbox/messageTraceFile-4.txt.gz (4MB) The extended format of the trace is: proc, ctxt, statIncrGCs, statTenures, statFullGCs, ioMSecs From this trace, the problem looks to be explainable as follows: - The time elapsed between two calls to WorldState>interCyclePause: is longer than 20ms. The cause for this is not the UI process using up a lot of time, but that there are 3 incr. GCs consuming 6 to 8ms each within each UI cycle. - Therefore, the UI process does not wait on the delay in WorldState>interCyclePause: - Because the UI main loop only does a yield (see Process class>>spawnNewProcess) the UI process therefore stays runnable all the time as there is no other process with p = 40. - Therefore, no process with p < 40 has a chance to be activated (only higher ones, which we find in the trace). This also explains why we see 100% CPU usage, but still the UI responds immediately. So, while the UI process runs, the incr. GCs use up most of the time and keep the UI process active. There is just not enough space requested to trigger a full GC. Now, why does moving the mouse make it run again? I have no idea... my guess is that the triggered behavior of a mouse move event somehow forces a full GC. In the trace we see that when the 107th full GC is done, there are much fewer incr. GCs later on. Hence, it is much more likely that the UI process pauses again. How could we fix this? ----------------------------------------- a) Simply increase the 20ms pause defined by MinCycleLapse (at least for production systems) or tweak the "pause logic". As a test I defined MinCycleLapse to be 40ms. I could not reproduce the problem anymore. b) In production, suspend the UI process and only turn it on again when you need it (we do this via HTTP). This should also improve performance a bit. At best this is a workaround. c) Tune the GC policies as they are far from optimal for today's systems (as John has suggested a couple of times). It seems, though, that this cannot guarantee to fix the problem but it should make it less likely to happen(?). I also took GC statistics while running the process (only part of it and not when it was stuck): - http://www.netstyle.ch/dropbox/gcStats-1.txt.gz: standard configuration (was very slow) - http://www.netstyle.ch/dropbox/gcStats-2.txt.gz: runActive with setGCBiasToGrowGCLimit: 16*1026*1024 d) ? I'd be interested in getting feedback on: - whether the explanation sounds plausible - whether the fix (e.g., a)) workes for other people that have this problem. - what may be a good fix Cheers, Adrian ___________________ Adrian Lienhard www.adrian-lienhard.ch

3 7

PNG translucent images in 3.4
by Thushar G R 31 Jul '07

31 Jul '07

Hi group, Does any one know where to find the changes for PNGReadWriter for 3.4 image with the fix for translucency?The translucent PNG image is not producing the desired result when opened in 3.4 image but in 3.9 its working.Wher can i find the changes so that my 3.4 image opens translucent images properly. Thanks in advance, G R Thushar

1 2

SqueakSource server image?
by Alejandro Martínez 31 Jul '07

31 Jul '07

hi all. anybody know where I can download a SqueakSource server image ready to use? the download link at http://wiki.squeak.org/squeak/5820 (Run a local SqueakSource server) is broken. cheers

3 2

KernelImage
by Pavel Krivanek 30 Jul '07

30 Jul '07

Hi all, here: http://comtalk.net/public/pub/KernelImage/KernelImage20070729.zip you can download latest KernelImage materials. It shows how to load MinimalMoprhic system simply using one file into a small kernel image (about 3000 methods). The package then includes scripts that tries to change license of non-free code in this small image. If the changes will be confirmed by you, we will have got the free kernel that can load the rest of current Squeak code. See readme.txt for more details. Cheers, -- Pavel

3 7

RE: Newbie question
by J J 30 Jul '07

30 Jul '07

Well, it's syntactic sugar. It looks pretty alien in Smalltalk, but other languages support multiple-assignment (e.g. "swap" {a. b} := {b. a}). Of course this gets tricky because if you have multiple assignment it seems natural to return multiple values from a function (e.g. {a. b} := a swapWith: b). That would probably require a big change to implement and you end up with something Smalltalk can already do other ways, and become (more) incompatible with other dialects.> To: squeak-dev(a)lists.squeakfoundation.org> From: blake(a)kingdomrpg.com> Date: Mon, 23 Jul 2007 12:58:25 -0700> Subject: Re: Newbie question> > On Mon, 23 Jul 2007 12:27:21 -0700, subbukk <subbukk(a)gmail.com> wrote:> > > On Monday 23 July 2007 11:39 pm, Bert Freudenberg wrote:> >> >> Now, at one point the compiler even supported this:> >>> >> {a. b} := {1. 2}> >>> >> which I found cool but was considered evil, even by those who> >> tolerate the braces ...> > What would this do? It looks like you're assigning a literal to another > literal?> _________________________________________________________________ Don't get caught with egg on your face. Play Chicktionary! http://club.live.com/chicktionary.aspx?icid=chick_wlmailtextlink

3 3

[GSOC] HTML Layout - Acid1 now CSS Layout Test Passes
by Todd Blanchard 30 Jul '07

30 Jul '07

Jerome is now officially a rock star in my book. See it here: http://jeromechauveau.free.fr/blogsoc/?p=19 Now if he can just get tables implemented before the end of summer, we will have a great start at a very credible HTML viewer. -Todd Blanchard

1 0

RE: Delay and Server reliability
by J J 30 Jul '07

30 Jul '07

It will be really nice if these changes can find their way back into the base image.> Date: Sat, 28 Jul 2007 12:00:18 -0700> From: andreas.raab(a)gmx.de> To: squeak-dev(a)lists.squeakfoundation.org> Subject: Re: Delay and Server reliability> > Adrian Lienhard wrote:> > Nice to see that these kind of production issues fixed.> > Andreas, if you have more patches laying around, let us know ;)> > You may want to check out the Croquet repositories[1][2]. We've posted > quite a few changes there that helped with general robustness issues. > The last round [3] had various interesting fixes some that helped with > reliability in general (like the handling of out-of-memory conditions).> > [1] http://hedgehog.software.umn.edu:8888/> [2] http://jabberwocky.croquetproject.org:8889/> [3] https://lists.duke.edu/sympa/arc/croquet-dev/2007-05/msg00035.html> > Cheers,> - Andreas> _________________________________________________________________ Don't get caught with egg on your face. Play Chicktionary! http://club.live.com/chicktionary.aspx?icid=chick_wlmailtextlink

1 0

KernelImage
by Jerome Peace 30 Jul '07

30 Jul '07

KernelImage Hi Pavel, Is there a place for reporting errors? Trying minamal morphic, I get the following: Sunday, July 29, 2007 8:39 PM For this one in a fresh minimal morphic 2 world menu>open>classic browser>browser image aborts with: dnu.txt saying: Error: Improper store into indexable object ------ miniimal morphic fares better. from a fresh minimal morphic. try opening a browser. (a selection box comes up) if package browser is selected the emergency evaluator actuates but then the browser and an error box come up. the second attempt to open a browser (no selection box this time) actually calls up a package browser. >From a fresh minimal morphic. Try opening a browser. (a selection box comes up) If browser is selected an error box comes up but no browser. ---- Back to the drawing board. Yours in curiosity and service, --Jerome Peace *** > > >Pavel Krivanek squeak1 at continentalbrno.cz >Sun Jul 29 19:46:30 UTC 2007 > >Hi all, > >here: http://comtalk.net/public/pub/KernelImage/KernelImage20070729.zip >you can download latest KernelImage materials. It shows how to load >MinimalMoprhic system simply using one file into a small kernel image >(about 3000 methods). The package then includes scripts that tries to >change license of non-free code in this small image. If the changes >will be confirmed by you, we will have got the free kernel that can >load the rest of current Squeak code. See readme.txt for more details. > >Cheers, >-- Pavel *** ____________________________________________________________________________________ Be a better Heartthrob. Get better relationship answers from someone who knows. Yahoo! Answers - Check it out. http://answers.yahoo.com/dir/?link=list&sid=396545433

2 1

Jump to page: