Squeak-dev January 2002

squeak-dev@lists.squeakfoundation.org

176 participants
612 discussions

Re: A Tool for managing modules
by Bergel Alexandre 31 Jan '02

31 Jan '02

> And I tried to import my Connectors change set (after changing all the > category names to be 'People-NedKonz-Connectors'). > > And I imported the change set. When it executes the postscript of the change > set, I get a walkback on a doesNotUnderstand:. I have tried to reproduce this behavior, but I couldn't. Where can I get your Connectors change ? > But this is the strange thing: the Package Browser shows that I've added the > new method to the class in question (this is a one-method extension to > MessageSend). > > It's in a module called > #(People NedKonz Connectors 'Squeak Language Core Objects delta') Strange name for a module, but anyway I will try to fix that. You can also try to perform : Repository importChangesFromFileNamed: 'MyGoodie.cs' intoModuleAt: #(Project MyGoodie) > And I see the method in the browser. > > But when I look for implementors, I see none! > > Am I missing something? > > How am I supposed to use the system if I can't extend existing classes? Which system do you refer (ModuleTool is simply a gui, a front-end to ModuleInstaller) ? Extending an existing class could be done easily, for example : MessageSend subclass: #Test instanceVariableNames: '' classVariableNames: '' module: #(People ab TestExtend) > Then I use the ModuleTool to look at what I just loaded. I ask to look at > 'People NedKonz', and I see all the modules. > > I click on 'People NedKonz Connectors' and all seems well. > > I click on 'People NedKonz Connectors Squeak Development Debugger delta' and > I get a walkback in ModuleTool>>updateStatePanel, because self > getModuleFromListIndex returned nil. > > Apparently the problem is that your code is trying to do this: > > Module fromPath: #('People' 'NedKonz' 'Connectors' 'Squeak' 'Development' > 'Debugger' 'delta') > > when it should be doing this: > > Module fromPath: #('People' 'NedKonz' 'Connectors' 'Squeak Development > Debugger delta') Okay okay, I will fix it. The problem is 'a b c' substrings => #(a b c) but 'a b c ''d e f''' substrings => #(a b c ''d e f'') as expected in fact Are you sure to want to use a such name ? Regards, Alexandre > > -- > Ned Konz > currently: Stanwood, WA > email: ned(a)bike-nomad.com > homepage: http://bike-nomad.com

2 1

[ANN] Packets for Squeak
by Stefan Krecher 31 Jan '02

31 Jan '02

Hello! A few month ago i announced the idea of a "squeak attack framework" - a nessus-like vuln.-scanner. I'm still working on that, but first of all I guess, that it's important to have a packet-injection and -capturing-engine for Squeak. So I programmed a little shared-library for Linux, to create Packets from Squeak (via pluggable primitives). The libnet-library is needed. On my homepage (http://www.krecher.de/) you can get the c-source of the library and some smalltalk-code, that show's how to use it. For example it is possible to send packets the following way: TCP new payload: 'howdy'; dstIp: '192.168.0.2'; send The next steps will be to improve the class-design an implement more protocols. I'm going to produce some more documentation, maybe on building plug prims. Then a packetcapturing-engine will be build, i think of using libpcap, but have no ideas of a possible design right now. Suggestions, testing, critiscism and help are very welcome. Best regards, Stefan -- PGP pub-key fingerprint = 2E77 C256 47B1 27B8 AE07 94BE D7A1 2A3C C716 2BE9 PGP public key: http://krecher.de/pubkey.asc "Wenn die Regierung dem Volk nicht traut, warum löst sie es dann nicht auf und wählt ein Neues?" - B. Brecht

6 8

Re: About the Registering File list
by ducasse 31 Jan '02

31 Jan '02

on 31/01/02 7:05 PM, Ned Konz at ned(a)bike-nomad.com wrote: > On Thursday 31 January 2002 03:31 am, Stephane Ducasse wrote: > >> So the file list only knows the classes and not the services which have >> been register. > > I'm in favor of letting the reader classes decide: > > * whether they're going to handle a specific file or not (not just by suffix!) > > * what services they want to provide for a specific file > > I'd suggest registering classes using file suffixes (as an optimization for > readers that care about suffixes), and then before popping up a menu, passing > the name of the selected file(s) to the registered reader classes. > > If you include '*' as a file suffix that can be registered, it would allow > readers to decide on a file-by-file basis. > > This is important for some files that may not have predictable file suffixes > (I'm thinking here of ZIP files, which may have suffixes like .zip .ZIP .exe > .EXE .pr .PR, etc...) and would benefit from being looked at (for magic > numbers, etc.). I was thinking about the same in fact. So I will try to use that. > That is, I'd have something like this (probably pseudo-code): > > ArchiveViewer class >>initialize > FileList registerReader: self forSuffixes: #( * ). > > ArchiveViewer class>>servicesForFile: aFileName > services := OrderedCollection new. > (aFileName isZipFile) ifTrue: [ > services add: (SimpleServiceEntry > provider: self > label: 'view zip' > selector: #viewZip:) > ]. > services add: (SimpleServiceEntry > provider: self > label: 'add to zip' > selector: #addFileToNewZip:) . > ^services > > and before popping up a menu, the FileList would call all the reader classes > that had expressed an interest in the given file suffix (plus all those who'd > registered '*'). The results would be added to the menu.

1 0

[BUG] 3.3 Browser lookup walkback
by Jim Benson 31 Jan '02

31 Jan '02

Open a System Browser. Do a 'find class ...'. Look for 'Project'. Select 'Project' from the menu. Squeak presents a "Message not understood" dialog for the method 'category'. Jim

2 1

re: Squeak VM, real-time I/O, and threading
by goran.hultgren＠bluefish.se 31 Jan '02

31 Jan '02

Howdy! Craig Latta <Craig.Latta(a)NetJam.ORG> wrote: > Hi-- > > > I agree with Goran completely on this issue. Subclassing is > > particularly in-appropriate for network protocols, since they should > > be stackable. > > I think there's some confusion over what artifact we're talking > about... Göran was talking about Flow's architecture (I think :), but Yep. Flow. I am not totally confused. :-) > Flow *does* use composition. A client or server (e.g., HTTPClient) uses > a stream (e.g., SocketStream), which uses a resource (e.g., TCPSocket). Oh yes - it is very nicely architected overall. It was just the Client and Server classes that disturbed me IIRC. > I guess you're taking issue with the way things are now in mainstram > Squeak? ("HTTPSocket", etc.) Well, no, but I agree - the same problem exists in the Sockets hierarchy in vanilla Squeak. But you fixed that part in Flow very nicely. > Göran... how would you change the proposed architecture of Flow out of > concern for composition? Eh, well, I don't have your code up in any image (nor the time right now) but I think I would try to rewrite Client and Server into some form of pluggable components or something. And then they should probably have some more descriptive name like... Gah... RequestDispatcher instead of Server? I believe (out of memory) that Server is a bit similar to ConnectionQueue/ConnectionHandler right? ConnectionDispatcher perhaps. Whatever. I want to instantiate one on a port and plug in some guy implementating a simple protocol getting fed a Stream. Then it can listen and create Sockets and Streams and Processes etc as long as my little guy gets called somewhere... :-) regards, Göran PS. Client and Server were probably added last (and perhaps with less thought) and they aren't big classes - but since that is probably where the beginner starts to look for examples they play an important role in showing "how yo do it" in Flow. DS PS. <CSOTD> "I don't think we need another snippet now that Modules have hit the town. God have mercy on our souls. ;-) Henrik - good work! DS

2 1

Dynamic system memory use
by Derrick Coetzee 31 Jan '02

31 Jan '02

I don't know a lot about this, so I might be misunderstanding the way it works, but it seems to me like Squeak takes a fixed size chunk of system memory on startup and then hands it to its own internal memory managers to do its own allocation scheme. If this is right, then this would have to be one of my primary complaints about Squeak; even the simplest, most stripped down program can take a large amount of memory, and, more importantly, a program that cannot anticipate beforehand how much memory it will need is unable to dynamically use only as much of the system memory as necessary (hurts runtime scalability). Is there a solution to this in Squeak or is this a design issue? -Derrick Coetzee

3 2

Re: About the Registering File list
by Duane Maxwell 31 Jan '02

31 Jan '02

Ned sed: > I'm in favor of letting the reader classes decide: > > * whether they're going to handle a specific file or not (not just by suffix!) I second this - there are a couple of cases where a simple single file suffix is insufficient. Not only are there multiple suffixes that could be a single file type, but there are potentially multiple intepretations of a particular suffix. For instance, a while back I wrote some code to handle fileIns of compressed changesets, which have the suffix '.cs.gz'. It would be nice if we could produce the fileIn menu item if that's the extension, but not if it's simply '.gz'. -- Duane

1 0

RE: [ANN] Packets for Squeak
by Withers, Robert 31 Jan '02

31 Jan '02

Awesome article, Luciano! I can't wait for the article on the TCP stack. It would be interesting to decode the raw Quake data, though. Or Age of Empires, for instance. Rob > -----Original Message----- > From: Luciano Notarfrancesco [mailto:core.lists.squeak@core-sdi.com] > Sent: Thursday, January 31, 2002 2:16 PM > To: squeak-dev(a)lists.squeakfoundation.org > Subject: Re: [ANN] Packets for Squeak > > > Stefan Krecher wrote: > > > Is that article available as plain-text, or do I realy have > to fetch that > > iso-image and burn it on cd? > > > It's available now in html format: > http://community.corest.com/~luciano/tcpip/quake/ > > > > Great work - do you have some sample code to show how to use it? > > > Are the tools you implemented freely available? > > > > > > I haven't published the rest of it yet. I'll let you know > when I do it. > > Cheers, > Luciano.- > > > --- for a personal reply use: Luciano Notarfrancesco > <luciano(a)corest.com> >

1 0

About the Registering File list
by Stephane Ducasse 31 Jan '02

31 Jan '02

Hi I'm fixing some of the problems with the file list. Now I have a question. Right now a reader registers in the following way: B3DScene class >>initialize FileList registerFileReader: self B3DScene class>>fileReaderServicesForSuffix: suffix ^(suffix = '3ds') | (suffix = '*') ifTrue: [ {SimpleServiceEntry provider: self label: 'Open 3DS file' selector: #open3DSFile:}] ifFalse: [#()] So the file list only knows the classes and not the services which have been register. The first question is then how to get some specific services A possible answer would be to call col := OrderedCollection new. FileList allReader do: [:each | res := each fileReaderServicesForSuffix: '*'. res isEmpty ifFalse: [col add: each ]]. col The problem is that every reader developers has to not forget to put the (suffix = '*') in the fileReaderServicesForSuffix method A possible solution would be to have a method such as B3DScene class>>registeredServices ^ {SimpleServiceEntry provider: self label: 'Open 3DS file' selector: #open3DSFile:} B3DScene class>>fileReaderServicesForSuffix: suffix ^(suffix = '3ds') | (suffix = '*') ifTrue: [ self registeredServices] ifFalse: [#()] The second question is should we register classes or services. With classes, when I'm looking for the services that fileIn for example I have to look for all the classes get their services select the right one. With services I would just goes over the services. Any ideas pros and cons. Stef

2 1

[modules] Reducing the image size by 38% and 50%
by Hannes Hirzel 31 Jan '02

31 Jan '02

Hi Henrik Gedenryd answered to my question 20 days ago: >There will be a grat deal of refactoring required before the module system >will work to shrink the image efficiently. Even with the module system, all >the horrible dependencies are still there and need to be removed. >So it will be a while before the old-style shrinking procedure will be >superseded. and yesterday he wrote >But what I really wonder is what made the image grow to 15+ megs from 9 or >so, it's nowhere near all code, I thought it was the "Welcome" projects but >that's just roughly a meg. It's really terrible that shrinking the image is >the blackest of Squeak arts. I did a test today how to reduce the image size of a Squeak3.3a-4664.image which is 16446kB. I did the following steps Select 'flaps...' and then 'destroy shard flaps'. Manually delete the sample projects. Execute Smalltalk zapAllOtherProjects This reduced the image size to 15326KB. Executing ChangeSet allInstancesDo: [:cs | cs zapHistory]. Smalltalk garbageCollect and saving the image showed that a new image size of 11839kB Deleting all the ChangeSets ChangeSorter removeChangeSetsNamedSuchThat: [:aName | (aName = 'New Changes') not] gave a new image size of 10170kB which is a total reduction of 38%. So the major part comes from the change sets not the example projects. I did a similar test some last week with a 4599 image and documented it on http://minnow.cc.gatech.edu/squeak/2182 (Smalltalk majorShrink) It gave a 50% reduction and I still had a running system which was fairly complete (for my needs). The page http://minnow.cc.gatech.edu/squeak/2182 contains copies from emails I collected from this list. Another test (not documented there) gave a resulting image (with Morphic) with just 4.5 MB which was fine. The basic conclusion is that you have to do your own version of "Smalltalk majorShrink". I hope this information helps. Hannes Hirzel P.S. Thank you, Henrik, for your work on modules! It's nice that it's now in the update stream.

2 1

← Newer
1
2
3
4
5
6
7
8
...
62
Older →

Jump to page: