February 2004 - cells-devel - mailman3.common-lisp.net

[cells-devel] Cello moving out
by Kenny Tilton 26 Feb '04

26 Feb '04

Cello is growing up (ported to Linux by Frank Goenninger) and is moving out to a place of its own: http://common-lisp.net/mailman/listinfo/cello-devel http://common-lisp.net/mailman/listinfo/cello-announce http://common-lisp.net/mailman/listinfo/cello-cvs Things should be quiet here until a planned massive re-write of Cells, which may not happen for a while, but if I run into a problem that traces back to the weakness I see in state propagation, look out. :) Sorry for all the spam. kenny -- http://tilton-technology.com Why Lisp? http://alu.cliki.net/RtL%20Highlight%20Film Your Project Here! http://alu.cliki.net/Industry%20Application

1 0

[cells-devel] Re: cells test breakloop
by Kenny Tilton 25 Feb '04

25 Feb '04

mikel evins wrote: > Kenny, > > When running cells::cv-test I hit a breakloop; the cells docs say that > this means something's wrong. > > Just for a sanity check (and because I'm considering whether to use > cells for some knowledge modeling in a personal project) I tried > running the tests from Lispworks, and broke in exactly the same way. > The output below is copied from OpenMCL 0.14.1, but is exactly the > same (modulo object numbers and break prompts) as that from Lispworks: Ah, that's one I had fixed for a while and then for some reason decided to break again. I am fixing it over here, but you can just either: 1. move on to try lesson-14 or starter; or 2. In c-break and c-calculate-and-set, change BREAK to ERROR Lispworks is right, ACL is wrong: break should land you in the debugger regardless of any errors. What I am doing in a few places during CV-TEST is deliberately coding usage errors with Cells to confirm user errors are signalled, but then I want the test to run to completion, so I trap the errors and actually signal a "test-failed" error if no error gets trapped. Under ACL the /real/ Cell error gets signalled with break and the test can trap it, but that is wrong. kenny

1 0

Re: Report on Celloglut and ACL 5.0.1 [was [cells-devel] Emailing: celloglut.zip
by Raistlin Magere 25 Feb '04

25 Feb '04

>>Loading Cloucell.lpr insteads gives a few errors still (not many at all): >> >>*Error Part1* >>; While compiling CELLS-TIME-REPORT in c:\dvx\cells\debug.lisp: >>Warning: EXCL::TIME-REPORT should be given exactly 9 arguments. It was >>given 8 arguments. >>Problem detected when processing >> (EXCL::TIME-REPORT CPU-GC-USER CPU-GC-SYS ...) >>inside (BLOCK CELLS-TIME-REPORT (TRC "cpu-gc-user" CPU-GC-USER) ...) >>inside (PROGN (BLOCK CELLS-TIME-REPORT (TRC "cpu-gc-user" CPU-GC-USER) >>...)) >> >> >>*Error Part2* >>in "structure-view" >>Error: No class named: EXCL::STRUCTURE-EFFECTIVE-SLOT-DEFINITION. >>[condition type: PROGRAM-ERROR] > >Sorry, I forgot about those. I'll nail those now. > >> >> >>*Error Part3* >>Cannot find c:\dvx\cello\lesson6.lisp. Will skip and continue compiling >>project. > >I am getting confused now, but I think the project to try is just plain >cello.lpr. new reorg underway (and last night's distro is definitely for >use with cello.lpr--you'll still have the above errors, but they can be >swept under the carpet). > Ok running cello.lpr (after having deletead all the fasl files) gives me only *Error Part1* i.e. the time-report error. Again lesson-14 loads fine just after loading. Is there any set of other tests (like lesson-14) that I can run (i.e. without having to read your code to figure out how to write my own tests - which I'll probably do at some other time)? raistlin _________________________________________________________________ Add photos to your messages with MSN 8. Get 2 months FREE*. http://join.msn.com/?page=features/featuredemail

1 0

Report on Celloglut and ACL 5.0.1 [was [cells-devel] Emailing: celloglut.zip]
by Raistlin Magere 25 Feb '04

25 Feb '04

Ok: Opengl-ffi.lpr loads fine and running lesson-14 shows the window with the fonts and moving text fine. Loading Cloucell.lpr insteads gives a few errors still (not many at all): *Error Part1* ; While compiling CELLS-TIME-REPORT in c:\dvx\cells\debug.lisp: Warning: EXCL::TIME-REPORT should be given exactly 9 arguments. It was given 8 arguments. Problem detected when processing (EXCL::TIME-REPORT CPU-GC-USER CPU-GC-SYS ...) inside (BLOCK CELLS-TIME-REPORT (TRC "cpu-gc-user" CPU-GC-USER) ...) inside (PROGN (BLOCK CELLS-TIME-REPORT (TRC "cpu-gc-user" CPU-GC-USER) ...)) *Error Part2* in "structure-view" Error: No class named: EXCL::STRUCTURE-EFFECTIVE-SLOT-DEFINITION. [condition type: PROGRAM-ERROR] *Error Part3* Cannot find c:\dvx\cello\lesson6.lisp. Will skip and continue compiling project. and that's it. Ah by the way in cloucell.lpr you still refer to "c:\\dev\\cello\\*whatever*" rather than "c:\\dvx\\cello\\*whatever*". _________________________________________________________________ Add photos to your e-mail with MSN 8. Get 2 months FREE*. http://join.msn.com/?page=features/featuredemail

2 1

[cells-devel] whoa, porters!!
by Kenny Tilton 25 Feb '04

25 Feb '04

Just killed about six hours on this: app ran fine when I built using the ACL project manager. Then I built with ASDF just to confirm the ASD files were complete. Suddenly the demo (the lighting panel) looked black and white, tho yellow check marks still appeared yellow. Quite a few hours into the show I discover the light positioner, if clicked, starts to bring things back to normal. But it starts as if it were uninitialized. Also, a few more clicks like this and we lose the whole IDE. Finally at hour six one of the monkeys accidentally changes the: (declaim (optimize...) at the top of the asd files to match the project manager defaults, and we are back in business. Under ACL/win32, that is, with less-than-optimal optimize settings. Which tells me... I got serious FFI declaration issues. I have been aware for some time that ACL can be very forgiving in this regard, and now I am guessing that goes away as the speed optimization gets cranked. So the good news is that I just need to crank the settings and then I should be able to debug my ffi work within ACL, my strongest environment. The upshot for porters is that their compilers may not be as helpful as ACL's, even with the safest and slowest settings. In fact, maybe I should test under Lispworks. That consistently broke on things ACL could handle, but it always turned out ACL had been letting me get away with murder. Well, ACL /is/ my strongest platform, and I have found a way it seems to make it less forgiving, so I may as well stay here until it all works with the toughest of settings, then see if LW is any less forgiving. There is no harm in trying if you manage to get a clean build and shared libs loaded and callbacks from C into Lisp working, but if the lighting panel comes up black and white (no apparent shading) and no spinning object (teapot)--bingo. btw, that apparent b&W is just bogus values for the ambience for the light. ambience is even lighting across a scene, so wicked bright ambience brings everything into high contrast. I will shortly put up CelloGlut.zip with this issue unresolved. The optimizations in the source /have/ been eased to make AllegroCL work when built via ASDF, but what your compiler does with what I am guessing is bad FFI work may still break things. kt -- http://tilton-technology.com Why Lisp? http://alu.cliki.net/RtL%20Highlight%20Film Your Project Here! http://alu.cliki.net/Industry%20Application

2 2

[cells-devel] cells-devel delivery disruption
by Kenny Tilton 25 Feb '04

25 Feb '04

sorry, everyone. cells-devel started (occasionally) demanding moderator approval, and I had forgotten any and all passwords. erik at cl.net bailed me out this am, and we should be ok until i get stupid again. :) kenny -- http://tilton-technology.com Why Lisp? http://alu.cliki.net/RtL%20Highlight%20Film Your Project Here! http://alu.cliki.net/Industry%20Application

1 0

[cells-devel] Emailing: celloglut.zip
by Kenny Tilton 25 Feb '04

25 Feb '04

Hopefully this gets thru. Too whipped to fire up the NT box where Yahoo's Java client for maintaining the web site works. I have mail into cl.net requesting my passwords be reset so I can start using the FTP there. cheers, kt -- http://tilton-technology.com Why Lisp? http://alu.cliki.net/RtL%20Highlight%20Film Your Project Here! http://alu.cliki.net/Industry%20Application

1 0

[cells-devel] God Bless Lispworks Again
by Kenny Tilton 25 Feb '04

25 Feb '04

Wow, I must have had to fix six things, but now the awesome lighting panel runs identically under ACL and LW on win32. And a problem I had even under ACL with nested windows is gone. Life is good. :) Stepping out for a well-earned pint, but I plan to upload before retiring so Europe has the best possible code to work from. kenny -- http://tilton-technology.com Why Lisp? http://alu.cliki.net/RtL%20Highlight%20Film Your Project Here! http://alu.cliki.net/Industry%20Application

1 0

[cells-devel] God bless Lispworks
by Kenny Tilton 25 Feb '04

25 Feb '04

Lispworks wasted no time going straight upside my head over an oddity caused by Freeglut history, I think: fonts are identified by enums 0, 1, 2 etc, but the parameter is declared in C as a pointer to a font! God bless static typing. AllegroCL I guess does things helpfully in this regard if speed/safety are set right, if not it does things insanely/silently. Lispworks just slapped me. The upsetting thing is that a couple of hours ago I happened to notice that some calls to the glut font functions had the ID wrapped in (make-ff-pointer...) and others did not. I know I was tired at that point because one of my cardinal rules is to fix anything like that whether or not I think it is related. Now Lispworks is yelling about something else. I do not think it should be, but I have to give LW the benefit of the doubt. kt -- http://tilton-technology.com Why Lisp? http://alu.cliki.net/RtL%20Highlight%20Film Your Project Here! http://alu.cliki.net/Industry%20Application

1 1

[cells-devel] Re: more portabello notes
by Kenny Tilton 25 Feb '04

25 Feb '04

Mikel, Hope you don't mind me replying on cells-devel. One never knows who might have something to contribute, and it might be good for team spirit knowing others are banging on this. I know it does my spirits good, esp. after a day like this (which suddenly just got a lot better--more separately). mikel evins wrote: > Just a brief status; I spent some time on it today. I'm hung on > getting the UFFI stuff to behave properly with OpenMCL 0.14.1, but > only because I have to do some other stuff. > > More progress soon. I am hopeful, because I hear MCL as a rule is strong on FFI, and methinks UFFI has long supported it. Thx for your efforts. kenny -- http://tilton-technology.com Why Lisp? http://alu.cliki.net/RtL%20Highlight%20Film Your Project Here! http://alu.cliki.net/Industry%20Application

1 0