Squeak-dev December 1998

squeak-dev@lists.squeakfoundation.org

125 participants
236 discussions

Re: Squeak: who's afraid of the big bad "crash"?
by Alan Kay 30 Dec '98

30 Dec '98

Tom -- I think this is a good approach. I've always thought of the "levels" as more on the same level, but with different kinds of fences around them ... Cheers, Alan At 11:05 PM -0000 12/28/98, Tom Morgan wrote: >SystemDictionary become: nil. > >I got curious when the fragment was posted, to think about >what exactly this means and what it does (in addition to >crashing the VM). > >In the responses to this post and in the discussion of >other impressively disruptive expressions, the idea of >'levels' kept coming up. Plainly, > > SystemDictionary become: nil. > >*means* more than a 'simple' assignment does. > >One of the points that has been made on the list is that even >'simple' assignment is/can be thought of as a meta operation -- >it is, in effect, the recompilation of the accessor for the >variable assigned to. > >Worthy of Lord Shiva, the operation: > > SystemDictionary become: nil. > >is spectacularly a meta-operation, since it destroys every >name in the image, and the possibility of ever naming >anything, at *any* meta level, ever again. > >One way of thinking about this could be to regard the assignment >as properly happening all the way up the reflective tower, at the >'top' of an infinite number of meta levels. > >This might be a clue about how to arrange an environment, >where these kinds of spectacular operations are possible, yet the >environment is safe for mere mortals. > >-As a mere mortal, I may make and revise the values in my image. >-As a demigod, I may make and destroy the names of mere objects. >-... >-... >-... >-As Shiva, destroyer of worlds, I may destroy the very ability >to name. >(It would be also nice to imagine the ability to >re-create the ability to name, within the confines of the >self same meta structure) >-- >I think that an important element of a re-do of the Smalltalk >meta structure should be an incorporation of the 'levels' of >meta operation in an explicit way. > >People seem to do live life with blurry use/mention distinctions >*most of the time*, so there should a light touch to the >explicitness of the meta levels -- there if you needed them, >in the background if not. > > ...Tom M > > > > >-- >+---Tom Morgan ------------------------------ (tmorgan(a)acm.org) -----+ >I Brooklyn Union Phone: 1 718 403 2427 I >I 1 MetroTech Center Fax: 1 718 488 1752 I >+---Brooklyn, New York 11201 USA ------------------------------------+

2 1

Midi Time Codes
by Steve Walker 29 Dec '98

29 Dec '98

I have been asked (by someone with much more knowledge of midi than I) to pose the question: Does Squeak (either with or without Siren) currently support the use of midi time codes for the synchronization of multiple midi devices? For example, can Squeak accept a time code signal from a digital tape unit and use that signal for synchronization ? Thanks, Steve Walker Saskatoon, Canada

1 0

Re: Update server for MorphicWrappers
by Gerardo Richarte 29 Dec '98

29 Dec '98

"Ted K." wrote: > Richie, > Let me see if I understand what you are trying to do: > I think I have not expressed myself right... The picture is this: - Not everybody has MorphicWrappers right now, so there is always a .ZIP file with the latest version (including latest updates) available in http://www.db.uba.ar/MathMorphs. - A lot of people is already using MW, but they have SOME version, and we want to be able to upgrade them to the latest one through an Update Server. the problem is: - Suppose you get the .ZIP file with the latest version. You'll still have Squeak Version 2.2 (for example). The next day, you want to check if there is a new update waiting for you, so you click Update from Server, and then you choose 'Morphic Wrappers Update Server', but you Squeak Version is 2.2, and below the line with '#MW2.2' there are a lot of upgrades. The problem is that, while your squeak is v2.2, the MorphicWrappers you have are newer than that Squeak... This problem arises when you want to maintain different threads of updates (for different projects). My idea is to check against the date of the update, not against a version... My intentin is not to change Squeak's version each time there is a new update for MW... Changer Bye! Richie++

1 1

Re: Window code issue
by vuletich＠ar.ibm.com 29 Dec '98

29 Dec '98

Hello In the OS/2 port of Squeak, I used a game-oriented API (DIVE) for transferring the pixels to the display. It uses hardware acceleration to transfer from a buffer that is allocated in video ram (if possible). So, the Squeak VM allocates one of such intermediate buffers, and transfer the pixels to it, doing the bit-endian conversion. This should be as optimized as posible. The pixels are then copied from the intermediate buffer to the actual display memory by the video card acceleration hardware, almost instantly. So, only one bit reversal is necessary. The results are quite satisfactory. The problem is that if you don't have enought video memory for the intermediate buffer, it is allocated from main memory, using it up, and slowing the display update. If you have such a game-oriented display routine for BeOS, give it a try. Juan Manuel Vuletich SW Export IGS- IBM Argentina Phone Nbr : 54-1-313-0014 ext. 5207 Tie Line : 840-5207 e-mail: vuletich(a)ar.ibm.com

1 0

Re: Squeak: who's afraid of the big bad "crash"?
by Michael S. Klein 29 Dec '98

29 Dec '98

A) I thought it was: Smalltalk become: nil. not: SystemDictionary become: nil. B) You may want to think about what sort of become: operations make sense. It is posible to do a become on both Smalltalk, and it's class, SystemDictionary, meaningfully and successfully. -- Mike Klein mklein(a)alumni.caltech.edu

1 0

Re: Squeak unstable on Linux ?
by Ian Piumarta 28 Dec '98

28 Dec '98

> Is morphic running correctly on anyone's (intel) linux box It seems to work for me -- but I haven't done anything "in anger" with it. Send me a reproducible problem and I'll see if I can figure out what's happening. (Assuming I can reproduce it. ;) Thanks! Ian

1 0

CrLfFileStream needs platform specific startup
by Pennell's 28 Dec '98

28 Dec '98

CrLfFileStream class guessDefaultLineEndConventions should be run during image startup so that it can determine the line end convention for the platform the image is currently running on. It seems safe to add this to add this to SystemDictionary processStartupList. ChangeSet below: ------------------------------------------------------ 'From Squeak 2.2 of Sept 23, 1998 on 28 December 1998 at 1:01:19 pm'! !CrLfFileStream class methodsFor: 'class initialization'! startUp self guessDefaultLineEndConvention! ! !SystemDictionary methodsFor: 'snapshot and quit' stamp: 'djp 12/28/1998 12:53'! processStartUpList "Call the startUp method on each object that needs to gracefully restart itself after a snapshot." DisplayScreen startUp. FileDirectory startUp. Cursor startUp. Smalltalk installLowSpaceWatcher. InputSensor startUp. ProcessorScheduler startUp. Delay startUp. CrLfFileStream startUp. Smalltalk startUp. Smalltalk isMorphic ifTrue: [World fullRepaintNeeded] ifFalse: [ControlManager startUp]. ! !

1 0

Re: Name resolution
by Marcel Weiher 28 Dec '98

28 Dec '98

> to give more background on my environment: my Mac is connected to a ISDN > router(using ethernet). I use network address translation to access my > provider, with internally non-routed addresses. > > The interesting thing is that I can reach the update server from netscape, > but, at the same time, cannot access them from within Squeak. The same > happens > when I use the Squeak browser: I can access sites using the IP number, but not > using their actual name. Interestingly it seems that if I have accessed a > site > once through IP number I can then use the actual name. However, I have to do > more tests to verify this. Are you using a proxy for Netscape? In that case, name-resolution for host-names in Netscape is not done by Netscape itself but by the proxy. In my MacOS-X-Server setup, I have the blue box (MacOS emulation running as a process within MOSXS) set up with an IP address not known to the outside world, and without a connection to DNS. Both IE and NS can access hosts by host-name via proxy or directly using IP addresses, but not directly using host-names. Marcel

2 1

Sound playback sample rate
by Richard L. Peskin 28 Dec '98

28 Dec '98

The latest version of the BeOS (R4) fixes the output rate of the audio stream to 44100Hz. What effect will this have on the System-Sound classes, i.e. what will I have to change (e.g. sqMacSound.c) to get the sound output to work correctly with this fixed sample rate? thanks, --dick peskin ================================= R. L. Peskin, Rutgers Univ. ; <peskin(a)caip.rutgers.edu>; <http://www.caip.rutgers.edu/~peskin> VT Phone (802) 824-4558 NJ Phone (732) 445-4208 "The corporate culture is concerned less with Occam's razor than his aftershave lotion."

1 0

Re: AlignmentMorph and TextMorph
by Pennell's 27 Dec '98

27 Dec '98

[snip] >Andrew's explanation probably answers this one. Incidentally, if the scores are 1000, 100, 10, and 1, then the 1 does not appear right-justified >in your example since there is (see TextMorph>>fit) a minimum size for TextMorphs. Using #contentsWrapped: instead of #contents: and >adding #rightFlush will correct this. Thanks - I read too much into the name #contentsWrapped and didn't look at the comments. #contents forwards to #contentsAsIs which disables autolayout. #contentsWrapped allows auto layout. Do you have any pointers on where to start exploring the layout code? >>The names are editable, the scores are not. (why?) >> >Well, they are editable in my system, but then I dug around a bit and found I had "fixed" this problem before. TextMorphs seem to think they >have an unlimited width and thus the first one on the line will answer true to #containsPoint: and the second one never sees it. My change is: >TextMorph>>containsPoint: >containsPoint: aPoint >(super containsPoint: aPoint) ifFalse: [^false]. >self startingIndex > text size ifTrue: >["make null text frame visible" >^ super containsPoint: aPoint]. >^ self paragraph containsPoint: aPoint Have you submitted this change to Squeak Central for inclusion in 2.3? -david

1 0

Jump to page: