<NeverDie>Is Guile scheme from the Guile in street fighter? <daviid>for info, Guile-Clutter 1.12.2.1 released <leedleLoo>It seems like the `map` executes on a single cpu core. Can I utilize multiple cores? <leedleLoo>holomorph, your're correct. par-map's working like a charm <leedleLoo>Is there any work being done to utilize the GPU (like openCL)? I know drivers are usually an issue. <mark_weaver>leedleLoo: no, guile has no facilities for using GPUs <leedleLoo>I'm interested in helping out that. I'll start cloning the guile repos and what not. Do you know if something like this has been attempted before? <leedleLoo>Or maybe any major roadblocks anyone can see? <mark_weaver>I think it's best to keep it as a separate library, at least for now. <mark_weaver>I guess we probably can't use "OpenCL" in the name, since Apple holds trademark rights to that name. <mark_weaver>my knowledge of OpenCL is weak, but from what I know, it's not obvious to me how best to integrate this with a high-level language like Guile. <mark_weaver>it might be a good idea to post your preliminary design ideas to guile-devel@gnu.org and request feedback <leedleLoo>I have some experience with OpenCL. From what I remember, finding libre drivers wasn't so easy <leedleLoo>I'll sign myself up with the mailing list then, thanks for the help <leedleLoo>Nouveau may have something. I'm not entirely sure though <mark_weaver>regarding the trademark issue: guile-opengl was formerly named guile-figl, before we determined that we could use the opengl trademark <mark_weaver>but it appears that work is being done on it, at least. <leedleLoo>mark_weaver: I can't find too much more info besides that matrix you linked to. I'll have to spend some time looking through the nouveau projects then. Thanks again <amz3>build a dht is not simple matter, especially with mutable keys <amz3>how I am supposed to make asymmetric signatures? anyone knows about this? <amz3>ACTION slighly off topic <lloda>any insight into 'Warning: Unwind-only `stack-overflow' exception; skipping pre-unwind handler.' when compiling master at 'BOOTSTRAP GUILEC ice-9/eval.go'? solutions? <wingo>lloda: could it be that your C stack limit is low? <wingo>i bet it's that you are compiling -O0 <wingo>and so the eval.c stack frames are taking too much space <CustosLimen>so I'm getting '(#f mutex not locked by current thread () #f)' <CustosLimen>the only thing accessing that mutex as far as I can see is bascially using with-mutex <CustosLimen>so unless with-mutex is broken this should not happen <CustosLimen>unless ofcourse guile has issues with pthread on solaris 10 <amz3>meh can't compile guile-gnunet <h0wl3vvd>guile specify order of evaluate arguments ? <davexunit>this person has been jumping around to all of the lisp related IRC channels that I belong to <vanila>maybe they're compiling a table of information <taylan>I think they're just the type of IRC user who joins to ask something and leaves again when it's answered. at least that's how they seemed to me on #emacs. <taylan>a fairly straightforward way to use IRC, yet seemingly strange :P <davexunit>it always seems like they were like "yup, this sucks, bye" <taylan>hahaha, they seemed polite to me <davexunit>they rubbed me the wrong way in #scheme awhile ago <CustosLimen>I don't get how this is possible though - the only thing accessing that mutex as far as I can see is bascially using with-mutex - so unless with-mutex is broken this should not happen - unless ofcourse guile has issues with pthread on solaris 10u10 <mark_weaver>CustosLimen: it might be a solaris-specific issue. if you can create a minimal self-contained example that demonstrates the problem, that would be helpful. then we could try to reproduce it on our systems and investigate. <CustosLimen>mark_weaver, if I could reproduce it fixing it would be easy ;) <CustosLimen>mark_weaver, was just checking if someone knows of solaris related thread issues off hand <mark_weaver>CustosLimen: not off hand, but none of the guile developers use solaris afaik. <CustosLimen>andm it looks like opencsw compiles without --with-threads <mark_weaver>it would be helpful to send a bug report to bug-guile@gnu.org, preferably with minimal self-contained code to demonstrate the problem, but even without that it would be useful. <CustosLimen>mark_weaver, yeah - still just fishing - once I have way to reproduce or tracked it down in another way will submit <mark_weaver>in the process where the problem occurs, are you using pure scheme code, or does the process include your own C code? are you using the dynamic FFI? <CustosLimen>mark_weaver, I dunno the code that well - but think its all scheme <CustosLimen>from stack traces it seems like problem is coming from scheme code anyway <lloda>wingo: it was indeed the stack, should have checked :-/ thx Limit was 8MB. <wingo>lloda: 8MB should be fine, but was it -O0 ? <lloda>no, it was -O3, then I tried -O2 and it didn't work either. <lloda>actually I had a question about the stack. I read about how the stack now grows automatically on the Scheme side, is it possible to use that from C instead of scm_gc_malloc() ? <mark_weaver>lloda: iiuc, only the VM stack grows automatically, not the C stack. <lloda>yes, I'm asking about the VM stack. The VM has to call into C sometimes, so it seems like it should be possible to manipulate the VM stack from the callee. Isn't it so? <mark_weaver>lloda: iiuc: since the VM is the only thing that allocates on the VM stack, it's the only place that should need to expand that stack, no? I don't understand why you'd want to grow the VM stack from anywhere else. <mark_weaver>my understanding is that callees to do manipulate the VM stack at all, although I confess I'm not 100% confident of that. wingo is the VM expert. <wingo>i think lloda wants to allocate data on the stack <lloda>well, precisely because the C stack cannot be grown or checked portably, so using a VLA is risky, but it bothers me to call scm_gc_malloc or scm_gc_malloc_pointerless for local data <wingo>the vm is not really designed for variable stack frame sizes, is the thing <wingo>it would be nice to support some kind of arena allocation but i think the vm stack is probably the wrong place to do it <wingo>my guess is that a better gc is the right solution :) <CustosLimen>mark_weaver, guix is nice - but I also dunno if it will work well on solaris <mark_weaver>CustosLimen: alas, it would have to be ported to Solaris. <mark_weaver>currently supported architecture+kernel combinations are: x86_64-linux, i686-linux, mips64el-linux, armhf-linux, with work-in-progress on GNU/Hurd. <CustosLimen>but actually to be fair I dunno if rvm works on solaris <mark_weaver>on the plus side, porting Guix to other architecture+kernel combinations requires only a modest amount of work, as long as GNU libc and GCC+binutils works well on it. I added the mips64el-linux and armhf-linux ports to Guix. The process is mostly automated, although there are typically minor problems that crop up along the way. <mark_weaver>but even so, I suppose this is not the answer you were looking for :-/ <davexunit>CustosLimen: rvm is terrible and best avoided. <davexunit>Guix is definitely our recommendation. it handles the *full* dependency graphs of any software. <davexunit>Guile, like Ruby, has some extensions that are implemented completely or partly in C, or interface with C libraries via the FFI <davexunit>Bundler can't handle the C dependencies, but Guix can. <turbopape>Hi guys, for this year's clojurecup, we implemented a little scheme using Clojure Macros : ***jmd` is now known as jmd
<hwpplayer1>I'm learning C , i see some java code and learning python too.Should i start with Lisp before Guile or just start with Guile and don't look back ? <davexunit>hwpplayer1: lisp is a family of languages, not a single one. <davexunit>the two major flavors are Scheme and Common Lisp. <mark_weaver>hwpplayer1: I would recommend the books "The Little Schemer" and "The Seasoned Schemer" <hwpplayer1>I like learning languages and it is not bad for me to be a Lisp like languages student <hwpplayer1>I'll look at Common Lisp , Scheme and Guile together <hwpplayer1>And i'll see what are the differences between them <hwpplayer1>davexunit : i couldn't understand do you mean language or something else ? <daviid>downloaded? I tought this book was 'purchase only' <OrangeShark>hwpplayer1: he means latest edition of the Little Schemer which is the 4th edition <daviid>paroneayea: wow cool! congrat. did not know this existed actually :) <OrangeShark>hwpplayer1: check the bottom of the site, it has a link to the source code <OrangeShark>it is hosted on savannah and I think the source code browsing uses cgit <paroneayea>davexunit: yeah, I was surprised that it really *was* no strings attached <paroneayea>davexunit: guixops might be a great project for it :) <davexunit>paroneayea: would be a hard sell to both my partner and employer to go work in SF for 3 months ;) <davexunit>I'm excited to see the cool stuff you're able to get done with mediagoblin during your time there! <daviid>paroneayea: me too: rewritting the all thing in guile :) <paroneayea>daviid: I kind of wish mediagoblin was in guile, but that might be hard to do at this point, and kind of a distraction from getting 1.0 out the door ;) <paroneayea>however, expect *future* federation web applications in guile :) <daviid>paroneayea: yeah, sure, I was joking, and know you'd like that too... <guile-guest7>Hi! Is there anything like (sum list), or do I have to write my own function? <paroneayea>an interactive fiction game where you learn lisp, written to run in the z-machine <hwpplayer1>Could you please show me a simple math example of Guile <vanila>You should read the free online book for learning scheme <hwpplayer1>i downloaded "The Little Schemer" and "The Seasoned Schemer" <mouldysammich>vanila: What is the free online book for learning scheme? Ive been looking to get a somewhat decent knowledge of scheme so i can adequetly develop things with Guile. At the moment i am just coasting off the Guile reference manual <OrangeShark>mouldysammich: I have a couple bookmarked, let me look for them <civodul>CustosLimen: (current-source-location) <paroneayea>oh cool I didn't know about (current-source-location) <CustosLimen>thanks civodul - I just realized it was right above current-filename ;/ <CustosLimen>was searching google and found some weird module you have to load to get ---FILE--- <CustosLimen>is there an alternative to (assq-ref something 'colour) ? <CustosLimen>is (string-concatenate "a" "b") shortest for for concat ? <mark_weaver>'string-concatenate' takes a single argument, which must be a list of strings. <mark_weaver>(string-concatenate <l>) is equivalent to (fold string-append "" <l>), but more efficient. <paroneayea>Playing with irregex and trying to write a Ledger parser for it because procrastivity