Squeak-dev April 2008

squeak-dev@lists.squeakfoundation.org

146 participants
187 discussions

[Job offer] Permanent Position for Seaside Developer in Switzerland
by Adrian Lienhard 21 Apr '08

21 Apr '08

Hi all, We are looking for a Web Application Developer to join our team. The new member will be participating in the conception, development, and maintenance of Cmsbox and custom business web applications. Work location is Bern, Switzerland. Cmsbox is a successful startup and has been awarded the "Master of Swiss Web 2008". We are developing a next-generation content management system. We are a young and ambitious team and work with Squeak/Seaside. For more information please visit http:// www.cmsbox.com and http://www.netstyle.ch (websites in German). Experience and qualifications: - Strong skills in software development with dynamic, object-oriented language, preferably Smalltalk - Good knowledge of web application development, including AJAX - Background in web technologies (XHTML, CSS, Apache, ...) and experience of Linux - Ability to work independently and quickly learn new technologies Candidates are invited to contact me via e-mail: adi(a)cmsbox.com. Cheers, Adrian CB | ADRIAN LIENHARD | CMSBOX · TERRASSENWEG 18 · CH-3012 BERN | TEL +41 31 356 42 52 · FAX +41 31 356 42 51 | WWW.CMSBOX.COM · MAIL(a)CMSBOX.COM "Never doubt that a small group of thoughtful, committed people can change the world. Indeed, it is the only thing that ever has." -- Margaret Mead

1 0

Error: ExternalUnixOSProcess not garbage-collected
by Winfried Jacobs 21 Apr '08

21 Apr '08

Hello list, I have a problem with ExternalUnixOSProcess objects. In my squeak program I create them a lot, but they do not get garbage-collected, and after 1200 of them are created, the program stops with an error: 'cannot create OS pipe' (error thrown in External pipe>>makePipe). The following script demonstrates what is going on: "### Script ###" 1500 timesRepeat: [ p := PipeableOSProcess command: 'ls'. "these commands don't make a difference ... :" "p finalize." "p release." "p processProxy finalize." "p close; closePipes." Transcript show: PipeableOSProcess allInstances size; show: '-'; show: ExternalUnixOSProcess allInstances size; space. ] . Smalltalk garbageCollect. "### End of Script ###" The Transcript shows that the number of PipeableOSProcess instances stays constant (more or less), while the number of ExternalUnixOSProcess instances goes up, and I have no idea how to get rid of them. After restarting the squeak image (save-quit-start) the ExternalUnixOSProcess objects are gone. The configuration: - vm-version: Squeak3.9alpha on a Ubuntu Linux - image version: Squeak 3.9 ("Ramon Leon's image") - 'OSProcess versionInformation' gives: - OSProcess version 4.3.6 - CommandShell version 4.3.2 (not installed in this image) Does anybody have an idea what's possibly wrong? Thank you in advance Winfried

2 2

Sake/Packages
by Keith Hodges 21 Apr '08

21 Apr '08

For those who are interested the anatomy of a task, AnyClass-c- #taskDoThis: aParam ^ SakeTask new: [ :task | "if a depended upon task fails then the process stops" task dependsOn: "list of tasks" { AnyClass taskDoThat. [ self isReady ]. (SakeTask class: 'SomeClass') exists. }. "block or list of tasks, the last item has to return true for the action to happen" "if a task fails, then the action is considered, not needed" task if: { (SakeTask class: 'SomeClass') exists }. "block or list of tasks" task action: { self doTheOther. [ "do something else" ]. ] there are a few wrinkles to be worked out but it has been working well for me. We await some file managing tasks as found in Make/Rake Keith

1 1

how to catch Exception in worker thread: LDAPLayer
by Stephen 20 Apr '08

20 Apr '08

I'm using the LDAPlayer library to authenticate username/password in a login form via a lookup to an LDAP database. If a wrong password is entered, then LDAPlayer throws an LDAPException in a worker thread but doesn't report the exception back to the calling thread. My problem is I am fairly new to Smalltalk and I don't know how to work around this situation. What I would like is to get the LDAPException back to my application so I can handle it as required. ------------------------------------------------- In case it helps, here are some code pieces: the call to do the authentication: [ conn := LDAPConnection to: host port: 389. [[ req := conn bindAs: userDN credentials: pwd] on: Error do:[:ex | Transcript show: ex messageText]. [result := req result] on: LDAPException do: [:ex |Transcript show: ex messageText]. ] ensure: [conn disconnect]. If wrong password is given, LDAPLayer throws an exception when evaluating "req result" but the Transcript message in my application is never called. In fact the workerthread (shown below) gets blocked by the exception. ----------------- LDAPLayer initiates a "WorkerThread" class with the following method start | process | running _ true. process _ [ [running] whileTrue: [ | element | element _ LDAPMessage newFrom: sockStream. Transcript show: '********* READ ONE OF: ', element asString; cr. self dispatchMessage: element. ] ] fork. ^ process -------------------- When the debug window comes up, the top method on the stack is "LDAPResult checkForExceptions", code shown below. checkForExceptions (resultCode = 0 or: [resultCode = 5] or: [resultCode = 6]) ifFalse: [ | ex | ex _ LDAPException newWithCode: resultCode. ex signal: errorMessage ] Note: confirmed commenting out the 'ex signal' line stops the debug window from displaying.

2 2

Whick Annoucement package?
by Edward Stow 20 Apr '08

20 Apr '08

Hi I know of three squeak announcements packages: Is their an 'official' package that is likely to be supported long term? I am also concerned about strong references causing my long running image to bloat, [3] seems to explicity address weak / strong references, I'm not sure of the others. The three are: [1] OmniBrowswer-Announcements - came pre installed in my Seaside image [2] Lukas Rengalli - MCHttpRepository location: 'http://source.lukas-renggli.ch/announcements' user: '' password: '' [3] SqueakSource - MCHttpRepository location: 'http://www.squeaksource.com/AXAnnouncements' user: '' password: '' Thanks -- Edward Stow

2 1

Re: [Webteam] Suggestion for Squeak Website
by Keith Hodges 19 Apr '08

19 Apr '08

Hello Michael, > > I'd be interested in hearing your thoughts on this approach. > > Cheers, > Michael > I have been using Bazaar to version control my images, I am just getting to grips with it, and I think it is fantastic, better than mercurial for managing squeak images. So I have set up a server that I can use, and make public at some point. My typical workflow involves, downloading the version I want, unzipping, copying, moving the version into my "Originals" folder, applying LPF, copying that moving that into my "LPF" folder, and then copy that to a working directory in which I start to load the packages that I need. In looking at the Bazaar user manual, it allows some interesting ways of organizing a reporistory of versions/branches. Here is a sample of my current repository structure. /squeak/3.10/.bzr /squeak/3.10/LPF/.bzr /squeak/3.10/LPF/pr_tools/.bzr /squeak/3.7/.bzr /squeak/3.8/.bzr /squeak/3.8.1/.bzr /squeak/3.9.1/.bzr /squeak/3.9.1/LPF/.bzr The directory /squeak/3.10/.bzr is a repository for the base version, which contains within it, the repository for the LPF image that was derived from 3.10 /Squeak/3.10/LPF/.bzr , which in turn contains a repository of my "pr_tools" working image, which was derived from the LPF version. A user can obtain a base 3.10 release, specifying a revision number or tag if need be, via: bzr checkout sftp://squeak.warwick.st/squeak/3.10 myworkingdirectory --lightweight to obtain the LPF derivative: bzr checkout sftp://squeak.warwick.st/squeak/3.10/LPF myworkingdirectory --lightweight to obtain a personal working branch for a new project based upon 3.10/LPF: bzr branch sftp://squeak.warwick.st/squeak/3.10/LPF mynewproject --lightweight to publish a the new project back to the shared repository: cd mynewpoject bzr push sftp://squeak.warwick.st/squeak/3.10/LPF/mynewproject Although I am not writing a book, I am wanting to simplify, a) the learning process, and b) the collaboration process. Although it is getting easier the process of enabling others to learn and come on board is still an uphill one. So my plan is to adjust my day to day working to use a number of "Public Images". I will have a single image as my working environment for 5 or 6 tools, and publish that as a version controlled deliverable. This deliverable will have the test environment for each package ready to run, and the documentation that is written can be version managed alongside or in some cases in the image itself. If this works, I can ensure that a properly configured image with of all of these tools is available for anyone to browse and try out. When it comes to collaboration on any one tool, there is a clearly defined and very easily obtainable test-bed for any new innovations. I am hoping that this deliverable, version managed in a public repository, will help me to address some of my own weaknesses. I often find myself publishing packages which unwittingly break things. This is fine, IFF you can adequately inform potential users, as to which versions are working correctly and which versions are still pre-release, which tests pass and which fail. Our current tools (mcz/universes) do not provide that part of the feedback loop and so it is often very difficult to inform our users exactly which combination of bits works as intended.... So my solution is to adopt these version controlled "reference images". With such a hierarchical version controlled server in place, it becomes much simpler to tell a user exactly which image they need for a particular tutorial. Furthermore, the documentation could be version managed together with an image as a coherent whole. just my 2p Keith

1 0

Re: [ANN] SqueakPeople Certifications
by Germán Arduino 19 Apr '08

19 Apr '08

Well, I developed some improvements and fixes: - Fixed the problem preventing ordering by alias. - Added a link to the SqueakPeople page of the person. - Added a history to hold on the evolution of the number of people by certification. Don't works automatically, I must update the data from time to time. - Added a report of such history data. The url of the live app is: http://squeakpeoplestats.seasidehosting.st/seaside/SqueakPeopleStats and the source code is available (MIT licensed of course) on SqueakSource. Let me know any comment. Cheers. 2008/1/28, Germán Arduino <garduino(a)gmail.com>: > > Thanks. > > Yes, I will do when time permit. > > Cheers. > > > 2008/1/27, Ron Teitelbaum <Ron(a)usmedrec.com>: > > > That is very nice. Any chance you could add a link to the name to the > > persons real squeakPeople account? > > > > Ron > > > > > -----Original Message----- > > > From: seaside-bounces(a)lists.squeakfoundation.org [mailto:seaside- > > > bounces(a)lists.squeakfoundation.org] On Behalf Of Germán Arduino > > > Sent: Sunday, January 27, 2008 7:25 AM > > > To: The general-purpose Squeak developers list > > > Cc: Seaside - general discussion > > > Subject: [Seaside] [ANN] SqueakPeople Certifications > > > > > > Hi: > > > > > > Trying to test the newest versions of Squeak (3.10) and Seaside (2.8) > > > I developed a tiny app to group the SqueakPeople by its certifications > > > using the page of SqueakPeople with this data, as data source. > > > > > > The app is running on seasidehosting, at this link: > > > > > > http://squeakpeoplestats.seasidehosting.st/seaside/SqueakPeopleStats > > > > > > Enjoy :) > > > > > > -- > > > Germán S. Arduino > > > http://www.arsol.biz > > > http://www.arsol.net > > > http://germanarduino.blogspot.com > > > _______________________________________________ > > > seaside mailing list > > > seaside(a)lists.squeakfoundation.org > > > http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside > > > > _______________________________________________ > > seaside mailing list > > seaside(a)lists.squeakfoundation.org > > http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside > > >

4 6

Squeak Etoys goes MIT license
by Hilaire Fernandes 19 Apr '08

19 Apr '08

It looks like it is done by now... http://blog.ofset.org/hilaire/index.php?post/2008/04/19/Squeak-Etoys-goes-M… -- http://blog.ofset.org/hilaire

1 0

message send logging VM
by John M McIntosh 19 Apr '08

19 Apr '08

Via a request from David Röthlisberger I've updated the special VM I had created to do message logging it's found via the experimental folder, in folder MessageLoggingVM on my .mac idisk or ftp from http://www.smalltalkconsulting.com/squeak.html Squeak 3.8.18beta3UXXX.app.zip This special VM based on the 3.8.18b3 carbon VM that logs the call stack on each message send. Because logging occurs at VM message send time it gives a very accurate picture of what the image is doing, and also generates gigabytes of data. Once you load the programming api change sets, usage is: Smalltalk setVMStatsTraceMessageSendLevels: 1. to set message send recording on. Note that you pass an integer so you can say Smalltalk setVMStatsTraceMessageSendLevels: 5. to set the number of levels in the call stack to print to 5. Smalltalk setVMStatsTraceMessageSendLevels: 0. turns the recording off, and closes the file, the file is always opened truncated writing. b) Or a kill signal to turn recording on/off and alter recording levels. kill -USR1 19278 {Pick a squeak VM process id} kill -USR1 19278 {increment the level by one, go to 2} kill -USR1 19278 {increment the level by one, go to 3} kill -USR2 19278 {turn recording off, and close the file} As a reminder David T Lewis also provided change sets to allow the interfacing to OSProcess if you wish to rebuild a VM with this feature and use OSProcess for more control. > VMStatsTraceControl-dtl > > Add VMStatsTraceControl to provide the signal handlers for SIGUSR1 and > SIGUSR1 controlling VM stats log. Sending a SIGUSR1 signal to the > Squeak > process increments the logging level, and SIGUSR2 disables logging and > closes the file. > > Requires OSProcess for the signal handling (but relevant methods can > be > copied into VMStatsTraceControl if you don't want to require OSP). The > signal handling primitive is in OSPP, currently distributed with the > Unix VM, or load OSProcessPlugin from SqueakMap. > > > VMStatsInterpreterChanges-dtl > > Interpreter changes to accompany VMStatsTraceControl-dtl. This removes > the SIGUSR1 and SIGUSR2 signal handlers from the interpreter so they > can be handled in the image. > > Load this after JMMRecordMsgSends and JMMRecordMsgSendInterface. > Also load VMStatsTraceControl-dtl. > > Interpreter changes: > - Removed #cCode: (interrupt handlers) from > Interpreter>>initVmStatsTraceMessageSendLevels. > - Added Add Interpreter>>primitiveGetVMStatsTraceMessageSendLevels. > - Added SystemDictionary>>getVMStatsTraceMessageSendLevels > - Fixed assignment bug in > Interpreter>>primitiveSetVMStatsTraceMessageSendLevels (change '=' > to ':='). -- = = = ======================================================================== John M. McIntosh <johnmci(a)smalltalkconsulting.com> Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com = = = ========================================================================

3 2

[ANN] First public version of rSTAdd-ons (Remote Smalltalk)
by Sebastian Sastre 19 Apr '08

19 Apr '08

Hi there, I've released rSTAdd-ons package. I've been working with rST and found some issues and improvements I needed which I was monkey-patching in this add ons package privately. As others here I've believe on the gains the open model can bring to us so I'm releasing all this in squeak source under MIT license. All tests are green. You can find it at rST repository. This add ons mainly add one feature and starts housekeeping which will be evaluated to selectively incorporate in main rST package in team work. If you use rST in your projects or for any reason you are willing to cooperate on maintaining and/or improving it please be my guest to write about it to coordinate efforts in that regard. For its future I've made some tests to prove we can improve its performance by about x10. Some discussion is needed about it to see the more convenient strategy for that. So today what's new about it? The more remarcable feature added by rSTAdd-ons package is in BlockContext. You can do things like: | salute | salute := 'hello world'. [1 to: 3 do:[:index| Transcript cr; show: salute, index asString] valueAt: '192.168.1.5' port: 9999. ...or this one: | salute remote | salute := 'hello world'. remote := 'Transcript(a)192.168.1.5' asLocalObject. [1 to: 3 do:[:index| remote cr; show: salute, index asString] valueAt: '192.168.1.5' port: 9999. (of course all that from other host than 192.168.1.5) That's was critical to be able to create new objects in the remote image from "data" in a client image. Also for updating objects. This allows to easily use an image as an ODB. If you add some disk support (like Sandstone) you have a pretty interesting persistence alternative. Specially for those who use Seaside or any scalability hungry architecture by allowing to prevent points of contention in the persistence layer. Other feature is about increasing security by using BoundedTcpListener so the broker accepts incoming connections to the specified ip and not from others. Also removed some dependecies on Microsoft Windows so in linux all tests can be green cheers, Sebastian Sastre

3 3

← Newer
1
...
5
6
7
8
9
10
11
...
19
Older →

Jump to page: