Squeak-dev July 1999

squeak-dev@lists.squeakfoundation.org

139 participants
367 discussions

Mac filetype setting problem
by Tim Rowledge 28 Jul '99

28 Jul '99

I downloaded the squeak2.4c image/changes to my iMac via Netscape. Won't run because they are 'Netscape Navigator' documents. What does one do to get around this irritating situation? tim -- Why Dogs Hate WIn95:- 4) Saliva-coated mouse gets mighty difficult to manuever. Tim Rowledge: rowledge(a)interval.com (w) +1 (650) 842-6110 (w) tim(a)sumeru.stanford.edu (h) <http://sumeru.stanford.edu/tim>

2 1

[BUG][FIX]SystemDictionary>logError:inContext:to: recursive failure
by Mark A. Schwenk 28 Jul '99

28 Jul '99

Here is a simple-minded fix that keeps Squeak from going off the deep when it attempts to report an error and the log debug stack to file preference has been set but the SqueakDebugLog.txt file cannot be opened. I noticed this because my Squeak directory was not writable by me. There are many other uses of FileStream that fail ungracefully under such conditions (such as filing out), though not in a recursive loop like the above failure. We should investigate these when we have added Exception support. -Mark Schwenk WellThot Inc. 'From Squeak 2.4c of May 10, 1999 on 28 July 1999 at 3:37:41 pm'! !SystemDictionary methodsFor: 'miscellaneous' stamp: 'mas 7/28/1999 15:05'! logError: errMsg inContext: aContext to: aFilename "Log the error message and a stack trace to the given file." | ff ctx | FileDirectory default deleteFileNamed: aFilename ifAbsent: []. (ff _ FileStream fileNamed: aFilename) ifNotNil: [ ff nextPutAll: errMsg; cr. Date today printOn: ff. ff space. Time now printOn: ff. ff cr. "Note: The following is an open-coded version of ContextPart>>stackOfSize: since this method may be called during a low space condition and we might run out of space for allocating the full stack." ctx _ aContext. [ctx == nil] whileFalse:[ ff print: ctx; cr. ctx _ ctx sender]. ff close]! !

1 0

[BUG]FileStream creation
by Tim Rowledge 28 Jul '99

28 Jul '99

The latest 2.5-to-be updates still don't include any fix for the problem of FileStream instance creation when a subclass is directly referenced. I don't care whether or not my suggested fix <http://swiki.gsug.org:8080/SQFIXES/107.html> is used, as long as some fix is made sometime. I'm sure a better fix could done by refactoring the hierarchy, but that's more work. Likewise http://swiki.gsug.org:8080/SQFIXES/110.html is still needed for Acorn. tim -- Useful random insult:- Always loses battles of wits because he's unarmed. Tim Rowledge: rowledge(a)interval.com (w) +1 (650) 842-6110 (w) tim(a)sumeru.stanford.edu (h) <http://sumeru.stanford.edu/tim>

1 0

Method hit counts
by Brian Sutton 28 Jul '99

28 Jul '99

After reading the discussions on method factoring and documentation, I would like to suggest an experiment: method hit counts. A minor enhancement to the Browser could count the number of times each method in Base Squeak is read by the Squeak user. Periodically, the counts could be added to a central database for all Squeak users worldwide. Methods with high hit counts would probably be 1. the most used methods 2. methods that are difficult to understand Even for a well-written method, at least one read is usually necessary. But if a method is read over and over, it is probably poorly written or inadequately documented. Thus, the Squeak community could find which methods need work. Like many Squeakers, I feel that big picture documentation is needed much more than API documentation. Maybe this experiment will identify groups of interacting classes and methods that need better documentation. In addition, assertions on issues like method factoring, for example Les Tyrrell's opinion from July 3, ----------------------------------------------------------------------- >There have been several times when I've called someone over to my >screen to show them some munged up collection of gargantuam methods, >only to be stymied by the fact that most of the methods are actually >quite small. The difference is that the methods I needed to understand >in order to do the repair were NOT the small ones, but the very >infrequent >BIG ones. In my experience, the bigger a method is, the more likely >it is to accumulate complaints. ----------------------------------------------------------------------- could actually be tested. Are bigger methods actually harder to understand, requiring more time in the Browser? Brian

1 0

RE: bug I'm seeing with forked process
by Jarvis, Robert P. 28 Jul '99

28 Jul '99

Are you running one of the OpenGL screen savers, like Pipes? Or maybe something like SETI@Home? Bob Jarvis The Timken Company > -----Original Message----- > From: Stephan B. Wessels [SMTP:stephan.wessels@sdrc.com] > Sent: Wednesday, July 28, 1999 2:00 PM > To: squeak(a)cs.uiuc.edu > Subject: bug I'm seeing with forked process > > BUG PROBLEM. > > I'm currently running Squeak 2.4b of April 23, 1999, Last update: 1209 on > Windows NT. > > The application uses a forked process which sits inside a loop waiting for > a > termination condition to exit. The program is processing a number of HTML > based > lists and creating a new dynamic output HTML document as a report. It's > been > very handy and was simple to write. > > The peculiar thing is it fails whenever I'm not at my desk. I suspect > it's > related to something NT is doing while running the screen saver. I get > the > following message repeating in my console: > CreateThread() failed (8) -- Not enough storage is available to > process > this command. > SetThreadPriority() failed (6) -- The handle is invalid. > > So what's the deal here? I'm pretty careful about forked processes and do > not > have any awareness of repeated forks in my code. The only thing is the > HTTP > stuff in the base image is probably doing some forked stuff. I wonder if > there's a resource exhaustion thingy I need to watch for? > > The thing that's so frustrating for me about this is that the application > was > something I crafted to make my own project management duties more > efficient. > Then a number of other managers have seen this and now before you know it > there's a whole bank of people how suddenly depend on good operation 24/7. > Of > course I didn't see a failure until it was out in front of everyone... > > Ideas? > > - Steve

3 2

Celeste saved my day
by Michael Rueger 28 Jul '99

28 Jul '99

Hi, just want to share a success story I had recently using Celeste. To make a long story short: a while ago I received an e-mail containing a single "." on a one line which caused every pop-client to refuse downloading this mail (and all the others waiting as well). The provider couldn't help me either (I rather suspect they didn't understand what the problem was ;-) ). Finally I remembered Celeste and it worked fine, I didn't even had to hack the pop code! Michael -- "To improve is to change, to be perfect is to change often." Winston Churchill +------------------------------------------------------------+ | Michael Rueger | | Phone: ++1 (818) 623 3283 Fax: ++1 (818) 623 3559 | +--------Michael.Rueger.-ND(a)online.disney.com----------------+

1 0

Re: bug I'm seeing with forked process
by Bob Arning 28 Jul '99

28 Jul '99

On Wed, 28 Jul 1999 13:59:54 -0400 "Stephan B. Wessels" <stephan.wessels(a)sdrc.com> wrote: >I'm currently running Squeak 2.4b of April 23, 1999, Last update: 1209 on >Windows NT. > >The application uses a forked process which sits inside a loop waiting for a >termination condition to exit. The program is processing a number of HTML based >lists and creating a new dynamic output HTML document as a report. It's been >very handy and was simple to write. > >The peculiar thing is it fails whenever I'm not at my desk. I suspect it's >related to something NT is doing while running the screen saver. I get the >following message repeating in my console: > CreateThread() failed (8) -- Not enough storage is available to process >this command. > SetThreadPriority() failed (6) -- The handle is invalid. > >So what's the deal here? I'm pretty careful about forked processes and do not >have any awareness of repeated forks in my code. The only thing is the HTTP >stuff in the base image is probably doing some forked stuff. I wonder if >there's a resource exhaustion thingy I need to watch for? Steve, A search of the list turned up this: =============================== 20 Apr Raab, Andreas RE: My 1st PWS Crash :-( This is a weird NT thing - for some reason NT does not want to create a thread for the socket. How long have you been running PWS without restart or reboot of the machine?! I would guess that restarting Squeak should solve this issue (if not, then a reboot of the machine might help ;-) In any case, getting the latest SPs is probably a good idea. Andreas -- +===== Andreas Raab ========= (andreasr(a)wdi.disney.com) ==+ | Walt Disney Imagineering Phone: +1 818 544 5016 I I Glendale, CA Fax: +1 818 544 4544 I +======< http://isgwww.cs.uni-magdeburg.de/~raab >========+ > ---------- > From: Vandendriessche, Frederik > Reply To: squeak(a)cs.uiuc.edu > Sent: Tuesday, April 20, 1999 4:54 AM > To: squeak(a)cs.uiuc.edu > Cc: recipient list not shown > Subject: My 1st PWS Crash :-( > > > > Anyone, does this ring a bell ? > > " > CreateThread() failed(8) - Not enough storage is available to process this > command > > SetThreadPriority() failed(6) - The handle is invalid > > " > > I got these lines in an erorbox, on the bottom of my Squeak window. > I run Squeak on a > Dell PC, x86 Family 6 Model 5 Stepping 2, > 130,480 KB Ram > Microzoft Windows NT 4.00.1381. > IE 5 5.00.0910.1309. > > I use the PWS for 4 months now, and like it verry much. > This is the first sight of this error, an error that occurred 'for no > reason'. > I mean, I was not doing anything special; just saving a page (max 300 > chars).... > > Does anyone know an answer, or a place where I can find it ? > Thanks. > > =============================== Cheers, Bob

1 0

Re: Ad: instance Variables...
by Peter Novak 28 Jul '99

28 Jul '99

As I said to Andreas Raab: > Hallo Andreas, > I am really sorry about it. In the evening (Central European Summer) > I realized that I spammed the list too much. I want to promisse, that > I'll never do it again. My excuse may be, that I was in hurry and didn't knew > if my experiments can be fullfilled. I needed to know if the thing I'm > trying, is possible to implement. I didn't realized that I am in Europe and > main part of members of the list lives in U.S. Nevermind. I hope that my > "guilt" is not so big :-|. > > Thanx for understanding... > > Peter Novak. I didn't want to disturb you so much I did. After all I solved that problem by adding instanceVariable to class MetaClass (it is easy) and created some protocol between Class and MetaClass. Now it's functional. Please excuse my mistake. Thanx Peter. -----Original Message----- From: Edward P Luwish <eluwish(a)uswest.com> To: squeak(a)cs.uiuc.edu <squeak(a)cs.uiuc.edu> Date: 28. júl 1999 17:05 Subject: Re: Ad: instance Variables... >> > >I think this has something to do with the Namespace project at Whitestein. Some time ago I sent out a proposal for implementing namespaces that basically added a namespace variable to every object in the system, and altered the behavior of the VM so that it would execute a method only in the namespace (or the super-namespace hierarchy) of the object that called it. While this does not implement everything in the Namespace proposal (such as resolving references to system and class-wide objects like class variables and pool dictionaries), it is the most radical >surgical procedure of the project and can not be accomplished without altering the VM. > >> >> As always, when you describe a problem on this list, please say something about what you are trying to do, not just how you trying to do it. For instance, if you simply want to add some attribute to any object (but not necessarily every one in the system), you might want to consider a mechanism similar to the dependents dictionary that makes every object *appear to* have an instance variable named dependents. > > Attachment converted: Anon:Peter Novak.vcf 8 (TEXT/R*ch) (00012164)

1 0

RE: Fwd: Quesa 0.0.2
by Lawson English 28 Jul '99

28 Jul '99

Raab, Andreas <Andreas.Raab(a)disney.com> >In fact, if you look closely at >Squeak-Alice then you'll notice that Squeak-Alice *is* a retained mode >framework - and a pretty good one at this because it allows you to script >arbitrary objects in the 3D world. That it is currently not at the same >speed as other retained mode frameworks has nothing at all to do with >Squeak-Alice. It is the immediate mode stuff which drops us in the >framerate >- everything in Squeak is currently run from software only, there is no >hardware accelleration at all. > Hmmm... What is the overhead of the retained-mode portion? My impression is that once you DO get hardware acceleration support, the fact that the retained-mode database is interpreted WILL be noticeable. In fact, I suspect that it already is noticeable in some cases. ------------------------------------------------------------------------- Lawson English. Squeak, snore, etc. Check out <http://www.squeak.org> -------------------------------------------------------------------------

2 1

RE: Fwd: Quesa 0.0.2
by Raab, Andreas 28 Jul '99

28 Jul '99

Jeff, Interfacing a retained mode framework from Squeak is - in my understanding - a very, very bad idea. The reason for this is that you're basically giving up all of the control to the framework (e.g., some stuff written in C++). While this is okay for certain applications that don't want to extend the framework it requires you to play by the rules not only of the underlying immediate mode framework (such as OpenGL, or immediate mode Direct3D, or Apple RAVE) but also by the rules of this retained mode framework [Jeff mentioned that they (e.g., the Alice team at CMU) had serious problems with some of the behavior of D3D when it comes to texture loads]. The question is: Do you gain enough by using the retained mode framework to give up the freedom you have in a Smalltalk system?! In my understanding the answer is clearly: No. The reason is that the retained mode framework is basically just a scene graph with a bunch of actions used for rendering, picking etc. All of this stuff is mapped to the immediate mode stuff anyways and we can do this *easily* in Squeak ourselves. In fact, if you look closely at Squeak-Alice then you'll notice that Squeak-Alice *is* a retained mode framework - and a pretty good one at this because it allows you to script arbitrary objects in the 3D world. That it is currently not at the same speed as other retained mode frameworks has nothing at all to do with Squeak-Alice. It is the immediate mode stuff which drops us in the framerate - everything in Squeak is currently run from software only, there is no hardware accelleration at all. So, what I think we really need to have is ultra-fast immediate mode rendering with a highly flexible retained mode framework behind it. The access to the immediate mode stuff will allow you to do real-time game stuff (I've never seen a game using any retained mode stuff) while the retained mode frameworks eases the use of 3D stuff for not so time critical applications (such as end-user scripting). Using Quesa or Open Inventor or any other retained mode framework only gives up the flexibility since you can hardly change anything without going to program C and thus I don't see any reason to go that way. Andreas PS. Since you're a lead Mac engineer: Is there any way to get HW accellerated rendering into off-screen bitmaps on a Mac?! If so, how do you have to setup things to make it work?! -- +===== Andreas Raab ========= (andreasr(a)wdi.disney.com) ==+ | Walt Disney Imagineering Phone: +1 818 544 5016 I I Glendale, CA Fax: +1 818 544 4544 I +======< http://isgwww.cs.uni-magdeburg.de/~raab >========+ > ---------- > From: Jeff Szuhay > Reply To: jeffs(a)pstnet.com > Sent: Tuesday, July 27, 1999 10:25 AM > To: squeak(a)cs.uiuc.edu > Cc: recipient list not shown > Subject: Re: Fwd: Quesa 0.0.2 > > At 10:33 AM -0700 7/27/99, Lawson English wrote: > >Not sure if anyone is really interested, but as I pointed out > >earlier, a collaboration between Squeak and Quesa 3D people would be > >of benefit to both camps. > > > I second this request. > > Remember QD3D/Quesa is not a replacement for OpenGL but would properly > exist on top if it and provide many things OpenGL does not. I can think > of several Squeak projects which would benefit from this (Alice for Squeak > is the first that comes to mind). > > 8-) > > --************************************************** > Jeff Szuhay <mailto:jeffs@pstnet.com> > Lead Macintosh Engineer voice: 412/271-5040 x227 > Psychology Software Tools <http://www.pstnet.com/> >

3 2

← Newer
1
2
3
4
5
6
7
8
9
...
37
Older →

Jump to page: