00:51:50 jauaor [n=araujo@gentoo/developer/araujo] has joined #ccl 01:01:04 Balooga [n=luke@208.87.19.36] has joined #ccl 01:01:30 Hi all. How stable is the cll/win32 1.3 build? 01:06:02 I ask as I'm trying to port lispbuilder-sdl to ccl, and things are not going too well. Stack frame errors erc. 01:16:41 What's a "stack frame error" ? 01:17:38 it was thrown up in the terminal. I'll get the exact error now, hold on. 01:18:10 trying to work through other package issues that clisp, sbcl, lispworks and allegro seem to have no problem with. 01:19:07 oops, thats a *features* thing, never mind. 01:20:33 "Exception on foreign stack 01:20:53 "Exception occurred while executing foreign code. 01:21:29 and you're in the kernel debugger ? 01:22:13 correct 01:23:15 If so, please do r (to show register values) and let me know the value of the flags register; I think it'll be called "eflags" or "%eflags". 01:23:58 %eflags = 0x00000647 01:24:40 and b should show a lisp backtrace. 01:25:40 I'm there. 01:25:55 mdc_mobile [n=mdc_mobi@ds9.entity.com] has joined #ccl 01:26:09 The call at the top of the backtrace is a cffi call to poll for SDL events. 01:27:15 And that would have called some callback that you defined ? 01:31:16 No, that's a foreign call taking an SDL_event struct. Returning 1 if there are pending events, or 0 if no events are available. If an event is avaialble, it is copied into the struct. 01:31:49 No callbacks involved at all. 01:32:00 Hmm. 01:32:21 OK then; you presumably need to figure out what's going on in that call. 01:32:32 indeed 01:33:25 There -is- a bug having to do with callbacks on x8632 that's fixed in svn, but that doesn't sound like the problem here. 01:35:20 It has to be something simple. I'll poke around. 02:54:52 bfulgham_ [n=brent@adsl-69-234-109-59.dsl.irvnca.pacbell.net] has joined #ccl 04:07:02 -!- Balooga [n=luke@208.87.19.36] has quit ["My damn controlling terminal disappeared!"] 04:36:28 -!- rme [n=rme@70.104.116.109] has quit [] 06:05:24 -!- bfulgham_ [n=brent@adsl-69-234-109-59.dsl.irvnca.pacbell.net] has quit [] 08:18:25 pem [n=pem@159.226.35.246] has joined #ccl 08:57:28 deylen [n=deylen@62.249.247.182] has joined #ccl 09:39:41 -!- deylen [n=deylen@62.249.247.182] has quit [Remote closed the connection] 10:30:59 -!- mdc_mobile [n=mdc_mobi@ds9.entity.com] has quit [] 13:24:54 segv [n=mb@ita4fw1.itasoftware.com] has joined #ccl 13:25:49 -!- lisppaste5 [n=lisppast@common-lisp.net] has quit ["Want lisppaste5 in your channel? Email lisppaste-requests AT common-lisp.net."] 13:26:21 lisppaste5 [n=lisppast@common-lisp.net] has joined #ccl 15:01:44 rme [n=rme@70.104.116.109] has joined #ccl 15:42:45 malsyned [n=malsyned@75.35.185.146] has joined #ccl 15:44:45 When I use save-application to create an image of my application, the resulting binary is like 20MB, even if the application is just "hello world". Can I unintern standard symbols and then run a GC in order to shrink the image before saving the it? Does that work? 15:46:07 15MB I mean. But still. 15:53:44 You're asking about something that's been called a "tree-shaker." CCL doesn't have one. 15:55:44 rme: Actually, the name "tree-shaker" sounds like it does more than what I'm asking. It sounds like a "tree-shaker" finds unused functions and gets rid of them. I'm just asking exactly what a tree-shaker does in the "get rid" phase to be sure a function is no longer in the image. 15:57:19 ccl function calls go through the symbol, so uninterning a symbol that's bound to a function that's called somewhere won't do much. 15:58:19 OK, but if the function isn't called somewhere, then uninterning and then forcing a gc would be enough to remove the function from the image? 15:58:40 I believe so. 16:00:01 OK cool. Does ccl export enough information to write a tree-shaker? Like, given a function can I get a list of all symbols it references? 16:00:55 It's possible to get that, but I don't know if it's exported. 16:01:42 I don't think I've ever found an open-source lisp that had such a feature, and it seems like a really useful one for deploying applications. 16:01:47 Another quirk is CLOS - the way clos works, it's hard for a clos class or method to be "unreachable", so you can't get rid of clos stuff this way. Which means every standard symbol referenced anywhere in clos is not going to get gc'ed. 16:02:01 There are only about 5 or 6 megabytes of functions in CCL, so I'm not sure that you will be able to win very much. (see output of CCL:HEAP-UTILIZATION). 16:03:19 gz, rme: Interesting! So if there's only 5 or 6 MB of functions, what's the other 10MB? 16:04:02 oh nevermind, I'll just read heap-utilization. 16:08:33 2MB of strings! wow. What is that, docstrings for all of the functions or something? 16:16:29 Various stuff. Docstrings, error messages, symbol pnames, etc. 16:17:56 -!- billstclair [n=billstcl@unaffiliated/billstclair] has quit [Read error: 60 (Operation timed out)] 16:20:34 When did CCL become so cross-platform? I remember a couple of years ago looking for a lisp that worked on mac, linux, and Windows and CCL was never mentioned as a possibility. 16:26:08 I think we got it working on Windows just earlier this year. The 1.3 release was made in April, and that included Windows ports. 16:27:46 -!- jauaor [n=araujo@gentoo/developer/araujo] has quit [] 16:34:22 Well, hopefully it'll take off because of it. I'm looking at it for switching to just because of its cross-platforminess. 17:32:04 Is there a way to get ccl to quit when I hit Ctrl+D? 17:32:35 (setq ccl:*quit-on-eof* t) 17:32:46 You can put in your init file. 17:33:03 rme: thanks! 17:38:56 Is rlwrap the only way to get readline support in ccl? 17:40:44 billstclair [n=billstcl@unaffiliated/billstclair] has joined #ccl 17:41:53 There's no built-in readline-like stuff, so something like rlwrap (or emacs, or the Cocoa-based IDE) is necessary for that. 17:44:19 anRch [n=markmill@nmd.sbx07279.burlima.wayport.net] has joined #ccl 17:44:31 OK, just checking. rlwrap works well enough for quick things, and slime for bigger things. 17:52:55 -!- anRch [n=markmill@nmd.sbx07279.burlima.wayport.net] has quit [] 18:21:11 -!- segv [n=mb@ita4fw1.itasoftware.com] has quit [Read error: 110 (Connection timed out)] 18:21:31 -!- malsyned [n=malsyned@75.35.185.146] has left #ccl 18:46:26 Balooga [n=luke@208.87.19.36] has joined #ccl 18:53:31 gbyers: Retrieved the latest version of CCL/wi32 from svn, and now lispbuilder-sdl works 100%. It is fast too. Faster than Lispworks, comparable to SBCL. 19:11:09 jajcloz [n=jaj@209.91.55.220] has joined #ccl 19:42:03 -!- jajcloz [n=jaj@209.91.55.220] has quit [Read error: 110 (Connection timed out)] 19:55:28 -!- Balooga [n=luke@208.87.19.36] has quit ["My damn controlling terminal disappeared!"] 20:32:32 milanj [n=milan@93.86.59.111] has joined #ccl 21:44:55 -!- alms [n=alms@146-115-42-237.c3-0.bkl-ubr1.sbo-bkl.ma.cable.rcn.com] has quit [] 22:49:33 alms [n=alms@146-115-42-237.c3-0.bkl-ubr1.sbo-bkl.ma.cable.rcn.com] has joined #ccl 23:03:45 -!- milanj [n=milan@93.86.59.111] has quit ["Leaving"]