Squeak-dev January 2005

squeak-dev@lists.squeakfoundation.org

134 participants
235 discussions

Re: TCP Sockets are unrealiable
by Chris Muller 11 Jan '05

11 Jan '05

> Hi, you might want to review the code at > http://www.smalltalkconsulting.com/html/OTNotes3.html Ma client server is based on this design and includes test scripts that send and receive as many thousands of requests/responses as you want to prove it.

1 0

Harvesting Party Saturday January 15th 16:00 GMT
by Ken Causey 11 Jan '05

11 Jan '05

Just a reminder that the next Harvesting Party will take place between 16:00 (4 PM) and 20:00 (8 PM) GMT Saturday January 15th, 2005. As usual we will meet in #squeak on irc.freenode.net. More information about the #squeak IRC channel can be found at http://people.squeakfoundation.org/article/7.html Information about the Harvesting Party idea can be found at http://people.squeakfoundation.org/article/29.html If you have the time please attend and work toward the future of Squeak. The next seven parties are scheduled for 20:00 (08:00 PM) GMT Wednesday January 19, 2005 00:00 (12:00 AM) GMT Monday January 24, 2005 04:00 (04:00 AM) GMT Friday January 28, 2005 08:00 (08:00 AM) GMT Tuesday February 01, 2005 12:00 (12:00 PM) GMT Saturday February 05, 2005 16:00 (04:00 PM) GMT Wednesday February 09, 2005 20:00 (08:00 PM) GMT Sunday February 13, 2005

1 0

[ENH] InformationSpace
by Brian Tabone 11 Jan '05

11 Jan '05

InformationSpace has been enhanced to add support for 1 Dimensional causal sets, such as those discussed in Stephan Wolfram's New Kind of Science. The user interface has not yet been updated to present this functionality to the user. The user must currently access the functionality directly from a workspace. New methods have been added: In class NumberSpace init1DWithXdepth: getBitAtX: A sample 1D Causal Ligature is provided, see class NKStriangle. For more information, please visit http://www.softcentral.com/informationspace/ Or download InformationSpace 0.6.0 from SqueakMap Questions and Comments are welcome -Brian Tabone

1 0

Re: TCP Sockets are unrealiable
by Frank Gerlach 11 Jan '05

11 Jan '05

FGSocketStream was missing in my posting. Here it is.. > Hello, > I am currently trying to develop an ORB for Squeak. Unfortunately, the > TCP Sockets seem to be very unreliable. Attached are a simple server > and a java client, which lock up Squeak after a couple of hundred > requests. The scenario is multi-threaded with 20 threads/Processes > running in parallel. > Maybe I am doing something which I shouldn't, so please comment my > code... > Thanks ! > > regards > > Frank > > > > > -- > +++ GMX - die erste Adresse für Mail, Message, More +++ > 1 GB Mailbox bereits in GMX FreeMail http://www.gmx.net/de/go/mail -- +++ GMX - die erste Adresse für Mail, Message, More +++ 1 GB Mailbox bereits in GMX FreeMail http://www.gmx.net/de/go/mail

3 2

[ENH] ImprovedBrowseIt
by Chris Muller 11 Jan '05

11 Jan '05

See http://lists.squeakfoundation.org/pipermail/squeak-dev/2005-January/087187.…

1 0

Re: what happened to alternativeBrowseIt in 3.9?
by Chris Muller 11 Jan '05

11 Jan '05

> Note that the purpose of #alternativeBrowseIt has nothing to do with > which browser is used. Note the comment in ParagraphEditor>>browseClassFromIt indicates a hierarchy browser is to be used. But I've noticed the ambiguity that I think you're alluding to. alternativeBrowseIt seems to do two things: 1) open with a Hierarchy browser instead of a package browser. 2) Be more friendly/responsive when the user requests to browse something that doesn't exactly match a class-name in the system. I think alternativeBrowseIt should toggle just between the two parts of #1. This allows the user/developer to focus primarily on the object-model or the package-model as he desires. (To have that function toggle between the PackageBrowser or the PackagePaneBrowser seems to contradict our favorite axiom that if two things are 99% the same we should make them the same or very different). For #2, how about we make it more user-friendly all the time, regardless of the alternativeBrowseIt setting. Today, when a new user browses "Int" without alternativeBrowseIt turned on the system does nothing; no response. When is this sort of behavior desirable? I've attached a changeSet which I think does better. It follows these rules in order of precedence: 1) If you type an exactly matching class name, it automatically browses that class. 2) If no exact match from #1, then it presents a list of the classes whose name partially matches what you've typed, even if the match is only in the middle. 3) If there is no match from either of the first two rules, then it looks for an exact-match of the capitalized Smalltalk-words inside the selection. For example, if you are looking at Interval>> from: startInteger to: stopInteger and you double click "startInteger" and browse it, the system brings up Integer. If you browse "aGlobalPoint" you get Point and if you browse "anOrderedCollection" you get a list to select between OrderedCollection or Collection. So the system provides a next-best-guess alternative based on Smalltalk naming conventions of prefixing with "a", "an" or some adjective. The third rule is very nice for fast browsing. I've been using it for years and have learned to type-imply my argument names so it will quickly open the correct browser. No need for Chuck when browsing my own code. Note, this changeset *only* changes the what-to-browse behavior; it does not affect *which* browser is used to present it. 3.9 is a for-developers release, I think this changeSet should be considered for inclusion. - Chris

1 0

Re: Why ProtoObject>>becomeForward: is not implemented
by Chris Muller 11 Jan '05

11 Jan '05

> Is there any reason why ProtoObject>>becomeForward: is not implemented? My understanding for the purpose of ProtoObject is to provide a slimmed-down inherited API for proxies and such. Therefore, I'll bet it is merely an oversight that becomeForward: is not included. But I'd also like to pose the inverse question as to why some methods are there. Mainly, doOnlyOnce: initialize rehash flag: ifNil: and its other three variants rearmOneShot If the whole point of ProtoObject is to have a minimal API, then aren't these rather extranneous?

2 1

Re: EventRecorderMorph
by Chris Muller 10 Jan '05

10 Jan '05

Hi Ned and Ralph, I'm not sure I understood all of Ned's post, but this part jumped out at me.. > However, I don't think that morphs get notified when their world is about to > become inactive, so you'd have to add an event notification from the project, > probably in Project>>finalExitActions... Another possibility may be the instance-based listener for leaving a world, someWorld when: #aboutToLeaveWorld send: #mySelector to: myObject which could be used to further notify its immediate submorphs. - Chris

1 0

REMINDER: Harvesting Party Tuesday January 11th 12:00 GMT
by Ken Causey 10 Jan '05

10 Jan '05

Just a reminder that the next Harvesting Party will take place between 12:00 (Noon) and 16:00 (4 PM) GMT Tuesday January 11th, 2005. As usual we will meet in #squeak on irc.freenode.net. More information about the #squeak IRC channel can be found at http://people.squeakfoundation.org/article/7.html Information about the Harvesting Party idea can be found at http://people.squeakfoundation.org/article/29.html If you have the time please attend and work toward the future of Squeak. The next seven parties are scheduled for 16:00 (04:00 PM) GMT Saturday January 15, 2005 20:00 (08:00 PM) GMT Wednesday January 19, 2005 00:00 (12:00 AM) GMT Monday January 24, 2005 04:00 (04:00 AM) GMT Friday January 28, 2005 08:00 (08:00 AM) GMT Tuesday February 01, 2005 12:00 (12:00 PM) GMT Saturday February 05, 2005 16:00 (04:00 PM) GMT Wednesday February 09, 2005

1 0

[UPDATES] 11 for 3.8 / 3.9
by Bert Freudenberg 10 Jan '05

10 Jan '05

Fixes collected from Mantis: "Change Set: DropService-bf Date: 21 September 2004 Author: Bert Freudenberg DefaultExternalDropHandler was using the name instead of the stream when performing its service, resulting in a DNU. Fixes bug #163" "Change Set: startupOrder-bf Date: 7 January 2005 Author: Bert Freudenberg Do byte reversal in ShortArrays earlier in the startup sequence. Fixes bug #184" "Change Set: MailPngFix-kfr Date: 5 November 2004 Author: Karl Ramberg Attached png would not show. Fixed." "Change Set: Embed-bf Date: 7 January 2005 Author: Andreas Raab / Jerome Peace Fix for bug #591 as suggested by Andreas Raab" "Change Set: DelayFix-dtl Date: 11 December 2004 Author: David T. Lewis The following will hang a Squeak image. The error cannot be caught and the image is left in an unrecoverable condition (at least I don't know how to recover it). (Delay forMilliseconds: 625/4) wait ==> BOOM The fix (provided) is to ensure that a Delays are created with integral values. " "Change Set: SharedPoolSpelling-tpr Date: 14 December 2004 Author: tim(a)sumeru.stanford.edu Mantis bug #666. A hopefully temporary fix for a correct-the-spelling problem when SharedPools are used. Since they are not dictionaries, the #keysDo: method is not really appropriate but it solves the problem pro-tempore" "Change Set: defaultDir-tak Date: 17 December 2004 Author: Takashi Yamamiya The argument of FileDirectory class >> setDefaultDirectory should be encoded text. A platform dependent string causes crash at start up. " "Change Set: AcornFDcheckName-75 Date: 23 December 2004 Author: tim(a)sumeru.stanford.edu Yet another attempt at catching and correcting filenames for RISC OS. This time we keep a class var with a mapping of all the allowable or convertable characters. This at least works ok with spaces and hardspaces" "Change Set: balloonCenter-wiz Date: 31 December 2004 Author: Jerome Peace Fix for Mantis bug 783: Balloon Morph text goes outside of balloons when lots of text. Large balloons such as that for the trash can are not centered in the balloon. adjustedCenter tried to fudge it. This fix returns the original center of the textMorph within the balloon by selecting the vertices that represent the morphs corners and finding their center. A perfect fit every time." "Change Set: keyboardCleanup-tpr Date: 5 January 2005 Author: tim(a)sumeru.stanford.edu Minor cleanup of HandMorph so it doesn't pillage EventSensor and grab the eventQueue. All the users of the modified code appear to keep working as before - menu navigation, typing etc" "Change Set: SharedQueuePeek-tpr Date: 5 January 2005 Author: tim(a)sumeru.stanford.edu The comment for SharedQueue>peek claims that the method will suspend a sender if the queue is empty. A simple test shows this to be crap. It is pretty obvious the comment/code was copied from ShareQueue>next and not well edited." - Bert -

2 1

← Newer
1
...
16
17
18
19
20
21
22
23
24
Older →

Jump to page: