Squeak-dev January 1998

squeak-dev@lists.squeakfoundation.org

81 participants
196 discussions

Re: #do: on MethodDictionaries
by Tim Olson 12 Jan '98

12 Jan '98

Alejandro wrote: >MethodDictionaries activates the Dictionary>>#do: method that is >implemented like: > >do: aBlock > super do: [:assoc| aBlock value: assoc value ] > >but... assoc is a CompiledMethod (not an association) and does not >undestand #value > >Is this a bug ? Yes, it looks like the do: protocol was missed, although associationsDo: and keysDo: were reimplemented for MethodDictionary. Those are the obvious enumeration protocol for MethodDictionary -- it is less obvious that do: is useful, which is probably why it was missed. The implementation should look something like: do: aBlock tally = 0 ifTrue: [^ self]. 1 to: self basicSize do: [:i | (self basicAt: i) == nil ifFalse: [aBlock value: (array at: i)]] -- tim

1 0

Re: On the security question...
by Tim Rowledge 12 Jan '98

12 Jan '98

On Sun 11 Jan, Christopher Oliver wrote: > I forgot to ask; does the snapshot primitive write objects which are > not live but not yet garbage collected to the image file? The primitive simply dumps object memory to a file; it's up to the Smalltalk code that does all the concommitant work to make sure it can be done properly. Typically, a garbage collect is done, files are closed, sockets shut down, lights turned out, note left out for the milkan etc just before the snapshot prim is called. An old 'classic' tech support problem is caused by unwary users arrang ing for snapshots to be done by a background process... so the snapshot file has something almost, but not quite, entirely useless in it, since the careful shutdown either doesn't get done or is effectively _un-done_. Bad news when the cu stomer is a major corporation and the person who suggested it is high profile 'expert' :-) tim -- "How many Trinoc does it take to change a lightbulb?" "Why do you want to know about our maintenance schedules? Are you planning to attack us in the dark?" Tim Rowledge: rowledge(a)interval.com (w) +1 (650) 856-7230 (w) tim(a)sumeru.stanford.edu (h) <http://sumeru.stanford.edu/tim>

2 1

Re: On the security question...
by Ian Piumarta 12 Jan '98

12 Jan '98

> I forgot to ask; does the snapshot primitive write objects which are > not live but not yet garbage collected to the image file? No. It does a full GC immediately before writing the image. Ian

1 0

Block ciphers in squeak
by Christopher Oliver 12 Jan '98

12 Jan '98

I'm in the process of trying to cruft up a portable secure terminal for system administration tasks. I'm working with Squeak since it pretty much runs the same no matter which platform I'm working with. Before I bend my back to the task of implementing a block cipher such as Square or Blowfish, I though I'd ask if anyone knows of existing work on this topic. All I found after a protracted session with Deja- News was an article by Smith suggesting implementation in assembler or C, but this defeats the goal of portability. I've already hacked up a PK algorithm that takes about 0.9 sec with a 1024 bit composite on a 133MHz Pentium using Piumarta's 1.23 Squeak. Thus, all I'm seeking is a sound block cipher and perhaps a hash. Regards, -- Christopher Oliver Traverse Communications Systems Coordinator 223 Grandview Pkwy, Suite 108 oliver(a)traverse.com Traverse City, Michigan, 49684 (define magic (lambda (f) (lambda (x) x))) (define (more-magic n) (lambda (f) (lambda (x) (f ((n f) x)))))

1 0

Equality test of Dictionaries
by Alejandro F. Reimondo 11 Jan '98

11 Jan '98

Equality test is implemented in Set as: -------------------------------------------------------- !Set methodsFor: 'testing' stamp: 'afr 1/10/98 17:56'! = aSet (aSet isKindOf: Set) ifFalse: [^ false]. self size = aSet size ifFalse: [^ false]. self do: [:each | (aSet includes: each) ifFalse: [^ false]]. ^ true! ! This causes a great degradation of perfomance when comparing a Set to itself because all elements are traversed. This perfomance cost is evident when using big Pool Dictionaries (that inherits from Set); each time you see a class using a big pool dictionary in a browser it waste a long time (determining poolDict names in Smalltalk #keyAtValue: ). I've solved this problem reimplementing #= as: ------------------------------------------------------------------ !Set methodsFor: 'testing' stamp: 'afr 1/10/98 17:56'! = aSet aSet == self ifTrue: [ ^true ]. "Modified because perfomance is degradated for big Dictionaries (like BIG pool dictionaries). Must be revised to conform X3J20 ANSI Smalltalk specification. (#= in AbstractDictionary conforms to Object protocol) afr 1/10/98 17:41" (aSet isKindOf: Set) ifFalse: [^ false]. self size = aSet size ifFalse: [^ false]. self do: [:each | (aSet includes: each) ifFalse: [^ false]]. ^ true! ! BUT, I don't know if it is the best aproach because X3J20 Specification determines that Dictionary protocol conforms to Object protocol; then equality test must be identity test! Comments on this topic will be appreciated. Ale. Alejandro F. Reimondo Feel free to visit Smalltalk User Group of Argentina (SUGAR) at http://www.sugarWeb.com

1 0

Etiquete for #startUp
by Alejandro F. Reimondo 11 Jan '98

11 Jan '98

What is the recommended mechanism to notify anObject (or aClass) than the image has just startedUp. I need to refresh some objects at system startUp and I don't know how to do it in a clean manner (without base method changes). In Visual Smalltalk when the image is startedUp an event is generated and it notifies all its dependents. How can I doit *clean* in Squeak? thanks. Ale. Alejandro F. Reimondo Feel free to visit Smalltalk User Group of Argentina (SUGAR) at http://www.sugarWeb.com

1 0

nil subclass: ... fails if evaluated more than once.
by Alejandro F. Reimondo 11 Jan '98

11 Jan '98

I have implemented a class like nil subclass: #SelfInitializingObject instanceVariableNames: 'block ' classVariableNames: '' poolDictionaries: '' category: 'Vsw310 Compatibility' ! nil implements #subclass:instanceVariableNames:... like trying to creating class as a child of Object and then changing the superclass (Object) to nil. When evaluating the expresion again it fails because it acts as a superclass change (to Object again) and the code for superclass change sends #removeSubclass: to the current superclass ( nil !) Please, correct this bug to silently avoid this situation. (afr 1/11/98 18:33) Thanks. Ale. Alejandro F. Reimondo Feel free to visit Smalltalk User Group of Argentina (SUGAR) at http://www.sugarWeb.com

1 0

#do: on MethodDictionaries
by Alejandro F. Reimondo 11 Jan '98

11 Jan '98

Is it posible that #do: does not work ok on MethodDictionaries ? (afr 1/1/98 16:57) aMethodDictionary do: [:each| each yourself ]. This code does not works. Oops. MethodDictionaries activates the Dictionary>>#do: method that is implemented like: do: aBlock super do: [:assoc| aBlock value: assoc value ] but... assoc is a CompiledMethod (not an association) and does not undestand #value Is this a bug ? Thanks. Ale. Alejandro F. Reimondo Feel free to visit Smalltalk User Group of Argentina (SUGAR) at http://www.sugarWeb.com

1 0

Re: New Squeak web page posted at UIUC
by Serg Koren 11 Jan '98

11 Jan '98

Hi, Feel free to put a link to my version of Squeak (1.24SK) on your site. http://www.VisualNewt.com/Mac/Squeak124.html Cheers, S ------------------------------------------------- VisualNewt Software: http://www.VisualNewt.com/ Maker of Newt'sPaper(tm) the Premier Newton(R) MessagePad(tm) News Reader. -------------------------------------------------

2 1

Makefile quesion: #include "sq.h"
by Brian Wildasinn 10 Jan '98

10 Jan '98

I've got a Makefile question for Squeak-1.23c2: Does anyone know why sqFilePrims.o, sqSoundPrims.o, sqUnixNetwork.o, and sqXWindow.c apparently aren't reading in their header file "sq.h" so that the proper symbols are geting defined during compile time, such as: #make ... sqFilePrims.o: Undefined symbol `_success" referenced from text segment. [snip] The sq.h file definitions don't seem to be getting into the above files, though the #include "sq.h" is at the top of each file. A Squeak Makefile is needed, which I'm trying to do, to work on a NetBSD-1.3 unix system which has not yet got a port of Squeak. Ian Piumarta <piumarta(a)prof.inria.fr> helped me out with sqUnixConfig.h by suggesting to include the gcc generated system info through gcc -E -dM c.c, which gave the m68k specifics. --although the /* remarks */ header notes in sqUnixConfig.h mentions lots of "HAS_SOME_FUNCTION" stuff that most config scripts seem to deal with--or can those be used as compiler flags, eg, gcc -DHAS_SOME_FUNCTION=1? Anyway, Ian mentioned that nobody yet has created a config script for Squeak. Is that still the case? By the way, I've been using Squeak on this same exact m68k cpu box running on top of MacOS 7.5. Under NetBSD unix I've been tinkering with gnusmalltalk-1.1.5 blox and stix browsers. The former is a more robust class heirarcy browser than the demo-like, but quicker stix browsers. I'd like to compare unix gst and Squeak once they successfully port."./configure" works on gst, whereas Makefile for Squeak needs the "#include sq.h" resolved. Any help is appreciated! Sincerely, Brian Wildasinn bwildasi(a)csulb.edu

1 0

Jump to page: