Squeak-dev March 2004

squeak-dev@lists.squeakfoundation.org

179 participants
629 discussions

Re: higher level languages
by Gary McGovern 01 Mar '04

01 Mar '04

Thanks for your feedback Lex. I get into that with HCI*. You almost read my mind about variables. You have to learn it !!!! How can that be overcome or should it ? People have their own fields of learning, maybe learning computing on top of that is inconvenient ??? Libraries is great,as I understand, that is one thing that makes Squeak a good meta language. *(diverging a lot, I argue with myself a bit about man-computer symbiosis.There is a symbiosis, there isn't a symbiosis :). Soon after reading it,thankfully, I read the Smalltalk paper, to serve the creative spirit, which sounds kinder. Man can do without computers, they just help him control his world) Thanks, Gary >"Gary McGovern" <zeppy(a)australia.edu> wrote: >> Can anyone say what the highest level languages will end up like. I like >> tiling but, I think there is something else. I think what I've used in >> Squeak might fall a bit short but I cannot quite put muy finger on it. >> >> I think drawing think drwing on the screen is the end - but how to script >> it , program it I think is still a problem. >> > >I don't know about the hard question. That is wonderful to think about, >however. Roughly, I think of (a) having lots of convenient *libraries* >around, and (b) I think of having my algorithms being effortless to >encode for someone who is experienced with the system. Oh, and (c), I >suppose there will be a major language where non-experienced people can >still struggle along and make useful things happen. This latter one >would include lots of friendly graphics and lots of drag and drop and >lots of feedback from the tool. But more specifically? I have no idea. > >A few notes, however. > >The available libraries in a language are an important topic that is >often overlooked. If you want to make a "simple" web server it could >require ten lines of code or a thousand, depending only on whether you >have a web server handy in your library. The library gives you the >words that your language can use. > >It is misguided to try and make all tasks accessible to completely >inexperienced people. I don't think you should need a college degree to >do most programming on computers. But, like driving a car or making >waffles, people should expect to spend some time on it before >internalizing it. This trend in HCI really bothers me. You shouldn't >have to study (much) to use a toaster, but a VCR really is complex >enough that it requires some thought and -- yes -- study of some sort to >even understand what it is doing. Likewise, for computer programming. >Make it easy for people who know what a variable is or who know that >tapes hold recorded video, but if you don't understand variables and >don't understand what a video is, you need to put in some study time. >Sure, build the tutorial into the interface if you want, but at some >point the user really needs to put the brain in gear. > >Finally, don't blow off text. Text is very powerful both for people and >for tool implementors, and so IMHO it should only be abandoned when you >are working on the extreme edges of non-studying users. People learn >Smalltalk syntax in a day or less, and Smalltalk syntax is enough to do >practically anything. Non-programmers do all kinds of things with >spreadsheets (or, maybe they are programmers now). So don't obsess over >the text. Heck, even Squeak's etoys are using tiles *to build >sentences*. > > >-Lex >

1 0

[BUG][FIX][TEST] BitBlt>>pixelAt:[put:]
by Andreas Raab 01 Mar '04

01 Mar '04

"Change Set: PixelAtBugFix Date: 1 March 2004 Author: Andreas Raab When using bit peekers or pokers across snapshots BitBlt may eventually attempt to write into the (compressed) bits of some bitmap, severely screwing up when it finally realizes that it needs to decompress the form. This CS forces unhibernation before the the bits are overwritten. Tests included"

1 0

Re: [FIX] FasterMouseOver-nk ( [sm][et][approved] )
by Andreas Raab 01 Mar '04

01 Mar '04

Hi Guys, Must have missed this post earlier on but I want to point out that this change will have side-effects. The current mechanism has been made so that even if morphs move they will be notified about mouse entry/leave events. E.g., if you have a little eToy that does something like: foo>>script1 [on: mousEnter] foo's color <- Color red. foo>>script2 [on: mouseLeave] foo's color <- Color green. foo>>script3 [ticking] foo forward: 5. foo turn: 5. then, upon having foo move under the cursor, the system will correctly invoke the event responses. With the changes proposed it no longer will. I can't say for sure what other implications this has but it's certainly an annoyance for eToys - in particular considering that #isUnderMouse will report true but no events are being generated. Cheers, - Andreas ----- Original Message ----- From: <marcus(a)ira.uka.de> To: <squeak-harvest(a)lists.squeakfoundation.org> Sent: Sunday, February 29, 2004 4:46 PM Subject: [FIX] FasterMouseOver-nk ( [sm][et][approved] ) > > No Conflicts. Changeset by Ned,reviewed by Tim and Daniel. Thus: > Approved. >

2 1

[BUG] Passing a LargeInteger to primSignal:atMilliseconds: hangs VM
by goran.krampe＠bluefish.se 01 Mar '04

01 Mar '04

I am reopening this BUG because I mistakenly closed it. The bug in the package that triggered it is fixed, but the bug itself - namely why the below hangs the VM is not fixed. /Göran ------------- (Delay forMilliseconds: 100 factorial) wait. The primitive fails (which is fine), a notifier comes up, and then the VM hangs (which isn't). This was uncovered by Radoslav Hodnicak due to a bug in the PostgreSQL Client (which was using Socket standardDeadline where it should have been using Socket standardTimeout - possibly a common gotcha for people converting to the Network Rewrite). Avi

1 0

[BUG] Code correction in PluggableTextMorph misaligned ( Hmmm, Nathanael? )
by goran.krampe＠bluefish.se 01 Mar '04

01 Mar '04

Ok, I have looked into this one in detail. The issue was triggered by "NS" (Nathanael?) in update 5700 (which Ken was unsure if it ever got properly reviewed btw) were he changed "self selectionAsStream" to "self selection string" in #evaluateSelection. The result of that change is that instead of passing a ReadStream with a position into the full text, it passes only the selected code into the Parser. That works fine until the Compiler starts to try to givinng feedback using instvar requestor. Those methods, like for example when it tries to correct a selector (this one was caught by Frank's patch) or when it reports a compilation error (not caught by Frank's patch) then thinks that the position it has are absolute and can be used in messages to requestor. But they are now relative. Also, I note that there is still a sender of #selectionAsStream, namely when you choose "debug it". Given Frank's patch that case is now instead broken. My conclusion: 1. Nathanael just did this little change as a side thought, otherwise he wouldn't have missed the other sender of #selectionAsStream. 2. If we indeed want the Parser to work like this then we need to fix quite a few places more - Frank's patch just isn't enough. Recommendation: Unless someone stands up and fixes this thoroughly so that there are no more senders of selectionAsStream ("debug it" works too), and syntax errrors are also fixed to work etcetc, I think we should back this change out and revert back to "self selectionAsStream". regards, Gran

1 0

Re: [ANN] BFAV 2.07 available on SqueakMap
by Brent Vukmer 01 Mar '04

01 Mar '04

I'm very pleased to announce that Frank Shearar and Tom Koenig have signed up to help maintain the BFAV! Please copy them on any bug reports or enhancement requests you have for the BFAV going forward. Thanks, Brent __________________________________ Do you Yahoo!? Get better spam protection with Yahoo! Mail. http://antispam.yahoo.com/tools

1 0

[ANN] BFAV 2.07 available on SqueakMap
by Brent Vukmer 01 Mar '04

01 Mar '04

This release features a number of helpful changes: -the details pane clears out while a post's body is being downloaded -The X/Y ratio in the title pane now shows the number of *groups* displayed to number of groups loaded -When clicking between tabs, the BFAV now remembers the last listing selection index for a tab. -the Easter egg in the title pane that triggered loading updates, is gone -file-handling for the zip downloads is more robust (should eliminate errors reported about 'latest.zip' Thanks to Frank Caggiano, Ken Causey and Frank Shearar for fixes and feedback. __________________________________ Do you Yahoo!? Get better spam protection with Yahoo! Mail. http://antispam.yahoo.com/tools

1 0

[BUG][FIX] Please keep those files closed...
by Andreas Raab 01 Mar '04

01 Mar '04

"Change Set: PleaseKeepFilesClosed Date: 1 March 2004 Author: Andreas Raab Two services (ChangeList & FileContentsBrowser) used open files from the file list as their arguments. Besides leaving dangling file pointers around (we should REALLY turn off finalization for files just so people get what they deserve) this may prevents anyone from actually renaming or deleting a file which is selected in a file list. The effect is particularly bad since it only shows for files that have a source file suffix - so if you have selected 'foo.bar' everything will work fine but as soon as you select 'foo.cs' or 'foo.st' things screw up. "

1 0

updated smallblog info
by Avi Bryant 01 Mar '04

01 Mar '04

If anyone's using SmallBlog, I've posted some info on how to get the slightly newer version I've been hacking on at http://minnow.cc.gatech.edu/squeak/3649 . One of the major advantages of this version is that if you configure it with your hostname, the links generated in the RSS feed will be absolute, and so will actually work. Also, accessing the front page and the RSS feed no longer involves HTTP redirects, which confused some aggregators. The configuration process for new blogs is also slightly friendlier now. Please let me know if you have any success or problems with my installation instructions. Cheers, Avi

1 0

Jump to page: