September 2009 - Pharo-dev - lists.pharo.org

Re: [Pharo-project] one click distribution

by Chris Kassopulo

On Fri, 11 Sep 2009 10:42:29 +0200, astares-Mmb7MZpHnFY wrote: > > now works on windows > > Any feedback from the other platforms? > > Bye > T. The -swapbtn option when executing Squeak causes problems with right button clicks on Linux. Symptoms include no context menus on click and halos around menus (if displayed). Maybe this option is deprecated by recent changes in mouse button handling? Chris

Sept. 13, 2009

[Pharo-project] I give up!!! Aka too much broken in the 64 bit world

by Schwab,Wilhelm K

Hello all, A couple of things that I really need to have working appear to be hopelessly broken on 64 bit linux, at least for now. After a lot of struggling this weekend, I finally decided to revert to 32 bits to be able to get some work done. I created a /home partition as part of the original installation, so it was not bad to do. At this point, it boots and is currently downloading a couple hundred updates. One interesting thing is that the remote desktop (tsclient) that I use to run LookOut!! on a box across town is working. Could I have a mix of 32 and 64 bit software? Maybe tsclient was 32 bit all along though; the file command might shed some light on that. If any of you have done this retreat and have some war stories that might be of use, please let me know. Bill

Sept. 13, 2009

[Pharo-project] Closure fixes - test & review

by Michael Roberts

I'm trying to integrate the latest batch of Closure fixes from Eliot (eem). These are listed on the tracker as 9 tasks - search under the closure tag. I'm looking for test & review of the changes. It is quite tricky and likely to kill your image. The merge / load into Pharo is challenging because we don't really want to just drop the versions into Pharo. It would back out a number of changes. Some trivial (comments, spacing) and some less so. Useful testing would include trying out your own application level code (e.g Seaside, etc) to see if anything breaks. useful review would be if you are familiar with changes/fixes you have put into Pharo, you can see if they are still there, do they still work. if you really know the compiler / execution machinery then of course... I'm only publishing a 1/3 of the changes at this point, so i'm not sure how much is expected to work in the middle, but Eliot was careful to split it up. 0.1- start with core image >= 10449 (you need the MC changes) 0.2- ScriptLoader new addRepository39ToAllPackages; addRepositorySqueakTrunkToAllPackages; addRepositoryTaskForcesToAllPackages I have published packages to the task forces area, but the others give you access to the source on squeak trunk and the ancestry on 3.9. Also, if you are really keen Eliot provided CS which are on each tracker item. There are pros and cons to trying to do this directly in MC or using CS. If you look at the tracker for the original package you can look at the merge directly from the Squeak trunk. The changes are in 8 parts, and tidyup/initialisation. Part: 1- <present> 2- Kernel-MikeRoberts-404 3- ---preload--- Tools-MikeRoberts.201 System-Support-MikeRoberts.88 ---actual--- Compiler-MikeRoberts.128 CompilerTests-MikeRoberts.20 KernelTests-MikeRoberts.149 (I have paused at 3 for this review because i break things after 4) Notes: 1- is already in Pharo (give or take a few things) 2- is a reasonable copy 3- I have started to leave changes out. Squeak appears to have recompiled everything at steps 3/4 but i've chosen to leave that till 9. partly because it breaks. This is why i have provided stubs of the closure initialization. thanks, Mike

Sept. 13, 2009

[Pharo-project] Polymorph - TabIndex and layoutFrames questions

by nullPointer

1- How i can change the #position or #extent of a Morph with a layoutFrame setted ? for example, i have a button "anchored" to top and right; if i set the #position: of that button the position donÂ´t change. Then i try override the #position: accessor of my button; first set #layoutframe: to nil, later do the set normally of #position: and later set again the old value of #layoutFrame: , for remain the original anchors. But donÂ´t works. Well, if I execute the code in a WorkSpace, line to line, then works! but donÂ´t if execute all code. I donÂ´t understand :| 2- How works the tab index mechanism in Polymorph? in some of my Views i have controls with unordered tab order in controls. How i can change the order? I donÂ´t see nothing for start. 3- Exists the possibility of change the position of cursor of a textbox? Regards, and thanks for the help -- View this message in context: http://n2.nabble.com/Polymorph-TabIndex-and-layoutFrames-questions-tp363804… Sent from the Pharo Smalltalk mailing list archive at Nabble.com.

Sept. 13, 2009

Re: [Pharo-project] getting rid of these boring questions and popup

by Schwab,Wilhelm K

With respect to the browsers in Pharo, I want to be told before I delete a class or a method; I do not want to be bothered before closing an inspector whose workspace I edited. I **DO** want to be told before closing a workspace with changed text. Just for the record :) As far as keyboard navigation in the browsers, that would be a fine idea. It probably won't surprise you to learn that Object Arts put some careful attention into that too. That in no way has to take the scrolled lists away from the rest of us (I spend a lot of time looking and pondering, for which clicking and scrolling is quite natural). The package pane could be ditched in favor of a right click and selection; the selected package could be shown in the title bar, since that space is already allocated and under used. Dolphin makes a repsectable case for showing a package pane, because it allows multiple selections therein. Even that can be handled by providing a filtered hierarcy browser. Bill -----Original Message----- From: pharo-project-bounces(a)lists.gforge.inria.fr [mailto:pharo-project-bounces@lists.gforge.inria.fr] On Behalf Of Philippe Marschall Sent: Sunday, September 13, 2009 2:47 PM To: pharo-project(a)lists.gforge.inria.fr Subject: Re: [Pharo-project] getting rid of these boring questions and popup Schwab,Wilhelm K wrote: > One of the things that set me off was "I think this whole 80ies style browser that's based on scrolling, clicking and popups doesn't cut it anymore and adding more tabs and buttons isn't gonna fix it." Imagine our collective foolishness using scrollbars and other familiar GUI metaphors, and telling a user they might be about to do something they might regret ;) Certainly prompts for confirmation can get out of hand and become counter-productive (that is afterall the subject of this thread). When I know where I want to go (class, method, ...) I am far faster there in Eclipse than in a Smalltalk browser. A big part of that is scrolling and clicking. Because these are unnecessary actions that just take time. Take the Firefox Awsome Bar as an example. Let's say you want to visit a site that you have visited before. You go to the Awsome Bar (shortcut Ctrl + L) and type part of the title. A list that is automatically filtered is displayed and you can select from there. Compare that with opening the bookmarks menu, selecting the right category, and then searching for the right entry. You don't notice how cool and useful this is until you switch to an other browser that doesn't have this like Safari. An other good example is the Spotlight Search in Mac OS. Staring an application, opening a document, ..., everything just a few key strokes away. No scrolling, no clicking. Code browsing should be like this and Eclipse is far closer to this than the Smalltalk browser. > However, you went on to say "Short anecdote, I our current project we don't ask the user for confirmation, ever." Taken all together, it sounds wreckless to me. Based on analysis of how the users used to old software. The pop ups just prevented them from getting work done. The didn't read them, really, often they just clicked them away before they were fully rendered. As soon as they want their pop ups back we put them back. So far nobody did. Cheers Philippe _______________________________________________ Pharo-project mailing list Pharo-project(a)lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

Sept. 13, 2009

Re: [Pharo-project] getting rid of these boring questions and popup

by Philippe Marschall

Schwab,Wilhelm K wrote: > One of the things that set me off was "I think this whole 80ies style browser that's based on scrolling, clicking and popups doesn't cut it anymore and adding more tabs and buttons isn't gonna fix it." Imagine our collective foolishness using scrollbars and other familiar GUI metaphors, and telling a user they might be about to do something they might regret ;) Certainly prompts for confirmation can get out of hand and become counter-productive (that is afterall the subject of this thread). When I know where I want to go (class, method, ...) I am far faster there in Eclipse than in a Smalltalk browser. A big part of that is scrolling and clicking. Because these are unnecessary actions that just take time. Take the Firefox Awsome Bar as an example. Let's say you want to visit a site that you have visited before. You go to the Awsome Bar (shortcut Ctrl + L) and type part of the title. A list that is automatically filtered is displayed and you can select from there. Compare that with opening the bookmarks menu, selecting the right category, and then searching for the right entry. You don't notice how cool and useful this is until you switch to an other browser that doesn't have this like Safari. An other good example is the Spotlight Search in Mac OS. Staring an application, opening a document, ..., everything just a few key strokes away. No scrolling, no clicking. Code browsing should be like this and Eclipse is far closer to this than the Smalltalk browser. > However, you went on to say "Short anecdote, I our current project we don't ask the user for confirmation, ever." Taken all together, it sounds wreckless to me. Based on analysis of how the users used to old software. The pop ups just prevented them from getting work done. The didn't read them, really, often they just clicked them away before they were fully rendered. As soon as they want their pop ups back we put them back. So far nobody did. Cheers Philippe

Sept. 13, 2009

Re: [Pharo-project] Network cleanup?

by Schwab,Wilhelm K

Stef, I wish I could answer that. I think some of the reports were false due to forgotten #initializeNetwork; that should not be happening to me now (my session manager should be doing it on startup), and I've tested with explicit initialization anyway. The name size primitive blows up, and since that is used as the sentinel for needing the old network, calling it with the old network in use (as happens) seems wrong. My thought is to herd all of the new and old code into separate classes so we (maybe just I) can keep it straight enough to figure out what is wrong. I just finished some exercise during which I realized that there is no reason to touch NetNameResolver; we can create something separate to replace it, and even faÃ§ade same from NetNameResolver for legacy code. Since the current code is broken, a complete faÃ§ade will be impossible, but the most common entry points could be provided. What do you think? Bill -----Original Message----- From: pharo-project-bounces(a)lists.gforge.inria.fr [mailto:pharo-project-bounces@lists.gforge.inria.fr] On Behalf Of StÃ©phane Ducasse Sent: Sunday, September 13, 2009 1:14 PM To: Pharo-project(a)lists.gforge.inria.fr Subject: Re: [Pharo-project] Network cleanup? would be good to fix that network problem. What is the status? Is it related to the vm or not? Stef On Sep 13, 2009, at 7:59 PM, Schwab,Wilhelm K wrote: > I'm still having trouble with DNS on Ubuntu, and it seems to be IPv6 > related. However, it gets very ugly to try to figure out what does > and does not (or is that should and should not?) depend on the > #useOldNetwork test. Would there be any objections to a split of > NetNameResolver to have have new and old subclasses, with the correct > class being chosen (or sub-instance created) on initialization? > > I was able to get around the troubles today by noting that (for now) I > have trouble iff I am on a Linux local host, but that trick will come > to an abrupt end soon. > > Bill > > > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project _______________________________________________ Pharo-project mailing list Pharo-project(a)lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

Sept. 13, 2009

Re: [Pharo-project] getting rid of these boring questions and popup

by csrabak＠bol.com.br

Em 11/09/2009 18:32, Schwab,Wilhelm K <bschwab(a)anest.ufl.edu> escreveu: > Yes, the popups should go away - far away :) A code mentor pane > with lint rule violations would be greatly appreciated, but the > modal messages are too much. Sometimes I want to save code that I > know will generate errors, with reasons ranging from "I'll let it > blow up and choose to implement it from the debugger" to "don't want > to lose this, so let's get it accepted before I do something > stupid." I think Bill here has an interesting point: there are lots of references on Smalltalk programming that advocate this kind of development. So it makes sense we could have some 'selection' mechanism on how Pharo should behave pertaining the the way this info is shown during the develpment session.

Sept. 13, 2009

Re: [Pharo-project] Network cleanup?

by StÃ©phane Ducasse

would be good to fix that network problem. What is the status? Is it related to the vm or not? Stef On Sep 13, 2009, at 7:59 PM, Schwab,Wilhelm K wrote: > I'm still having trouble with DNS on Ubuntu, and it seems to be IPv6 > related. However, it gets very ugly to try to figure out what does > and does not (or is that should and should not?) depend on the > #useOldNetwork test. Would there be any objections to a split of > NetNameResolver to have have new and old subclasses, with the > correct class being chosen (or sub-instance created) on > initialization? > > I was able to get around the troubles today by noting that (for now) > I have trouble iff I am on a Linux local host, but that trick will > come to an abrupt end soon. > > Bill > > > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

Sept. 13, 2009

[Pharo-project] Network cleanup?

by Schwab,Wilhelm K

I'm still having trouble with DNS on Ubuntu, and it seems to be IPv6 related. However, it gets very ugly to try to figure out what does and does not (or is that should and should not?) depend on the #useOldNetwork test. Would there be any objections to a split of NetNameResolver to have have new and old subclasses, with the correct class being chosen (or sub-instance created) on initialization? I was able to get around the troubles today by noting that (for now) I have trouble iff I am on a Linux local host, but that trick will come to an abrupt end soon. Bill

Sept. 13, 2009