Squeak-dev October 2000

squeak-dev@lists.squeakfoundation.org

112 participants
424 discussions

Re: Re: Transcript is not thread-safe???
by jchludzinski＠worldkey.net 28 Oct '00

28 Oct '00

Ned, >Hmm, why a class variable? It should be an instance variable, in >case you want to make, for instance, Transcript2 (another >TranscriptStream). I wondered about that too but at the time I had a prototype to demo and I didn't have the time to debate if this was optimal or not. >All this does is serialize #endEntry, which only works for calls >to #endEntry, or atomic calls to #show: (which also calls >#endEntry) FROM THREADS WITH THE SAME PRIORITY. All any critical sections (of code) do is guarantee mutual excution - i.e., serial execution. >Morphic was not designed, as far as I can tell, to deal with >multiple threads. (For instance, what happens to the Transcript if >you're manually editing or scrolling while logging from other >higher-priority threads? Resizing? Selecting?) I hope there's NOTHING inherent in Morphic which makes it ill suited for thread-safety??? Isn't Java's Swing (set) thread - safe??? Since Squeak support Processes (heck, there are many concurrently executing at any one time in Squeak - I monitor the count), shouldn't the TranscriptStream be made thread-safe? Shouldn't Morphic be made thread-safe? ---John

1 0

Transcript is not thread-safe???
by jchludzinski＠worldkey.net 28 Oct '00

28 Oct '00

Ned Konz wrote: >The Transcript is not thread-safe, for one thing; for another, you >make the UI run every time you send stuff to the Transcript (at >least when you do an endEntry). It was suggested by someone on this mailing list months ago that I add & change TranscriptStream as follows, 1) add AccessSema as a class variable: WriteStream subclass: #TranscriptStream instanceVariableNames: '' classVariableNames: 'AccessSema ' poolDictionaries: '' category: 'Collections-Streams' 2) change #endEntry to: endEntry self semaphore critical: [self changed: #appendEntry. self reset] and 3) add: semaphore ^ AccessSema ifNil: [AccessSema _ Semaphore forMutualExclusion] This seems to have work (at least for me!). ---John

1 0

campSmalltalk@OOPSLA pictures
by Donald MacQueen 28 Oct '00

28 Oct '00

http://users.erols.com/dmacq/ -- Donald Donald M. MacQueen Registered Smalltalk Bigot, [Esq.]

1 0

Re: UPDATE ON: Squeak becomes (almost) autistic???
by Ned Konz 28 Oct '00

28 Oct '00

jchludzinski(a)worldkey.net wrote: > One additional thing to consider: the Processes (at least for a > while after Squeak becomes autistic) continue to write to the > Transcript. And then eventually that stops. Ack! don't make non-UI threads write to the Transcript!! Figure out some other way to display things... The Transcript is not thread-safe, for one thing; for another, you make the UI run every time you send stuff to the Transcript (at least when you do an endEntry). Best bet is to just put debugging info into a collection. Make a circular queue, perhaps, so that it doesn't grow infinitely large. You can inspect the collection from an Inspector. -- Ned Konz currently: Stanwood, WA email: ned(a)bike-nomad.com homepage: http://bike-nomad.com

1 1

Re: Re: UPDATE ON: Squeak becomes (almost) autistic???
by jchludzinski＠worldkey.net 28 Oct '00

28 Oct '00

Bob, >You have Processes (plural?) running the models and putting data >into the queue. Then you have a Process taking it out and sending >it through a socket. And squeak gets real slow. Then you kill all >the model processes and everything is back to normal speed. >Right so far? Yes, you've got it. >So I would try: - leave the model processes running and kill the >sending process. How is it now? - take a look at the model >processes. What are they doing? The models ALSO have a Process associated them. The same as the Daemon #receivingRun and #sendingRun - an infinite loop with a 1 second Delay at the bottom. They are also created via a #fork. >Do they have Delays periodically? See above. >Do they wait on some other resource? Nope, the simple take Events of the EventQueue (a Heap) and execute them - part of a Model's logic (the Event handlers). >Do they yield? Nope, they use Delays (1 second). >Are they at the same priority as the UI and the sending process? - >have you tried this with the sending and receiving images on >separate machines yet? I assume they have a priority of 4 if I read the #fork logic correctly.

1 0

Re: Swiki.net vs. Comanche ?
by Bijan Parsia 28 Oct '00

28 Oct '00

On Fri, 27 Oct 2000, Bob Houston wrote: [snip] > I, too, am very interested in building web applications using Comanche. > But before I reinvent anything, I wanted to know if any of the advanced > features in swiki.net are available to other Squeakers. Below are the > features I need to build: > > 1. User login and session management (cookie or non-cookie based) This shouldn't be too hard. > 2. Secure Socket Layer support There is some camp smalltalk stuff toward this end, including using and external SSL thingy, IIRC. > 3. Database access and/or SOAP support > > The first 2 items on my list appear to be implemented in Swiki.net; and > I've seen discussions of the 3rd item on the Squeak swiki. SOAP is a Camp Smalltalk project. It's being built on the VisualWorks XML parser, I believe. I've ported the VW parser to Squeak (though the port and the parser have a ways to go), but no one, to my knowledge, has looked at bring the SOAP stuff over. (I'll be happy to help on the parser side, but I *saw* the team working on SOAP at CSt2 and I'd rather drive furniture tacks into my gums than work on it!) > Who else is building web apps with Squeak? Please drop me a note if > you'd like to discuss ideas. Folks on the PWS list. Try there too. >Also, I'd be glad to discuss the pros/cons > of building a pure object oriented framework for hosting web applications. Bolot wrote a bit on this for the Networking Squeak chapter and has a few papers discussing this. Interestingly enough, ComSwiki really isn't a Comanche based application (and is becoming less so). Indeed, it's arguably not *Smalltalk* based in that it's not a *Smalltalk* framework. It has an inheritence framework for Swikis that's sorta prototypey, and a page/book/shelve dispatch system (for "actions"). And it's file based :) Cheers, Bijan Parsia.

1 0

Re: Swiki.net vs. Comanche ?
by Jimmie Houchin 28 Oct '00

28 Oct '00

I too am interested in web development with Squeak/Commanche. I would like to see what all options and tools are available for web develpment. What does Swiki.net use that is available to the community? It would be nice to see a transcript, slides or something from Stephen's demonstration at OOPSLA. Tuesday 4:15 pm - 5:00 pm, Wednesday 4:15 pm - 5:00 pm 18. Implementing NetUnify.com Using Squeak as an App Server Stephen Pair, NetUnify.com Any and all information appreciated. Thanks, Jimmie Houchin Bob Houston wrote: > > I'm interested in learning more about the implementation of the > swiki.net site. If this info has been previously posted, please direct > me to its location. Swiki.net is the most impressive extension to the > swiki/comanche code I've seen. > > I, too, am very interested in building web applications using Comanche. > But before I reinvent anything, I wanted to know if any of the advanced > features in swiki.net are available to other Squeakers. Below are the > features I need to build: > > 1. User login and session management (cookie or non-cookie based) > 2. Secure Socket Layer support > 3. Database access and/or SOAP support > > The first 2 items on my list appear to be implemented in Swiki.net; and > I've seen discussions of the 3rd item on the Squeak swiki. > > Who else is building web apps with Squeak? Please drop me a note if > you'd like to discuss ideas. Also, I'd be glad to discuss the pros/cons > of building a pure object oriented framework for hosting web applications. > > Thanks, > Bob Houston

1 0

Test: Is the list up?
by Ned Konz 27 Oct '00

27 Oct '00

Just a test message; haven't seen anything that I've posted in some time. -- Ned Konz currently: Stanwood, WA email: ned(a)bike-nomad.com homepage: http://bike-nomad.com

3 2

MORE INFO: Squeak becomes (almost) autistic???
by jchludzinski＠worldkey.net 27 Oct '00

27 Oct '00

Currently (until I get it working) the 2 Squeaks that are running are on the SAME Win2K box. Eventually Linux is the target, but Win2K is easier for now to work with. ---John

1 0

AGAIN MORE: Squeak becomes (almost) autistic???
by jchludzinski＠worldkey.net 27 Oct '00

27 Oct '00

Here is what my timing calls do: delay1 _ Delay forSeconds: 1. [true] whileTrue: [ "store tank proxy or event data into a ByteArray (bar)" t _ Time millisecondsToRun:[ osProcessToSocket valuesDo: "currently ONLY one object in osProcessToSocket" [:v | socket _ Socket newTCP. socket connectTo: v ip port: v port. socket waitForConnectionUntil: Socket standardDeadline. bytesSent _ socket sendData: bar. socket closeAndDestroy]]. Transcript cr; show: '--->> sendingRun millisecondsToRun: ' , t printString. delay1 wait] delay1 _ Delay forSeconds: 1. [true] whileTrue: [ t _ Time millisecondsToRun:[ ss _ self cQ getConnectionOrNil. ss ifNotNil: [(ss waitForDataUntil: Socket standardDeadline) ifFalse: [self error: 'receive timeout']. bytesReceived _ ss receiveDataInto: bar. bytesReceived = 20 ifTrue: ["create event from info in bar"]. bytesReceived = 24 ifTrue: ["update tank proxy from info in bar"]. ss closeAndDestroy]]. Transcript cr; show:'--->> receivingRun millisecondsToRun: ' , t printString. delay1 wait] The 'delay1 wait' is intentional. I thought it might be less taxing for Squeak than 'Processor yield'. Is there ANYTHING that I'm doing that is in the slightest way suspicious??? ---John

1 0

← Newer
1
...
4
5
6
7
8
9
10
...
43
Older →

Jump to page: