<lloda>rest parameters go into a list and if you don't give them then the list will be empty. Is this what you mean? <lloda>#:optional parameters that are not set will be SCM_UNDEFINED in C. <madsy>lloda: Ah, okay. So an empty list then <madsy>Heh.. strange. (equal? #f '()) returns #f here <ArneBab>mark_weaver: another small status update: wisp now supports [] as well as {}. It does not do consistency checks on them, though: stuff like "{[}]" is pushed through unchanged. <ArneBab>(just wanted to share the late-night results - bbl) <lloda>madsy: in [Scheme] that's #f. <madsy>lloda: Ahh, right. I sometimes confuse Lisp semantics with Scheme's <madsy>So to test for the empty list in Guile, I have to use SCM_EOL *dje42 is opening a Guinness before checking in guile support to gdb. <lloda>madsy: there's scm_null_p iirc <lloda>or maybe better C Function: int scm_is_null (SCM x) <davexunit>how does one deal with configure variables like pkgdatadir in guile programs? I have a file or two that I want guile-2d to load like a default font. <davexunit>dsmith: well I'm trying to figure out what needs to be done. <davexunit>the value of pkgdatadir depends on the system <davexunit>but at runtime my library needs to know where to load things from. <nalaginrut>Anyone played Go? Is it possible goroutine was implemented with something similar with delimiter-continuation? Or they are different? <nalaginrut>Maybe it's a wrong opinion to mention delimiter-continuation for a total different language? <nalaginrut>well, goroutine is M:N, seems more complicated than I can imagine... <civodul>i have yet to try the latest Guile-SSH <artyom-poptsov>So... Does it work as expected? :-) Or you have not tried it yet? <artyom-poptsov>I did some testing of Guile-SSH on my machines before the release, but I suppose that some bugs can be lurking there. <civodul>i started uprading it in Guix, but there were test failures, possibly due to the build env. <civodul>i should be able to look into it sometime this week <ArneBab_>mark_weaver: thanks ☺ the new website is actually a static Mercurial repository created with my staticsite extension ☺ <ArneBab_>mark_weaver: I was sick of big hosters who try to create a walled garden around DVCS, so I created a static repository exporter which allows serving a Mercurial repository from any commodity server <ArneBab_>mark_weaver: so anyone can have that design easily <ArneBab_>(the design is mostly ported from stevelosh.com — I loved the site when I first saw it, and in the last weeks I discovered that Steve had licensed his whole site under MIT) <ArneBab_>mark_weaver: from what I see, all that’s still missing at the moment are multiline comments <civodul>oh, Guile 2.0's birthday is in 6 days! <wingo>civodul: want to send the mail? :) <civodul>not sure i'll have energy to hack, but email is fine ;-) <wingo>i wonder if i could do a little native compiler for a subset of things that i can completely inline and that don't need gc... *civodul prepares the messages <taylanub>Already been a whole year since the last birthday ? O_o <add^_>wow, read the backlog or whatever I should call it. Yeah <add^_>civodul: I dunno if I'll have either the time or energy for a potluck, but I can try! <add^_>I'll try making something more useful this time.. <add^_>Unlike my crappy unfinished guile-haiku <add^_>Not sure if I want to make it only for guile though, want to make it runnable between different scheme implementations <add^_>Maybe I should just skip that for now? <add^_>and go ahead and make something for just guile anyway.. <add^_>wingo, civodul: is there a way to have _standard_ r6rs keywords for define? <wingo>no idea what you mean, sorry <add^_>I thought define* was nonstandard.. <add^_>I'll try and figure it out.. <taylanub>add^_: R6RS doesn't have keyword symbols, does it / <add^_> I was thinking of #:key in the lambda forms <add^_>but that's probably an extra feature for guile <taylanub>add^_: Yes, that's a Guile extension, along with `define*'. Though it looks like SRFI-88 and SRFI-89 standardize both; you just need to (read-set! keywords 'postfix) and perhaps be careful to conform to some restrictions that SRFI-88 and 89 have which Guile doesn't, or vice versa. <lloda>wingo: on top of ra0 1a51415615fbe0f0fce4cd97437b846b82894d9a <lloda>uniform-vector? -> f64vector? <wingo>lloda: go ahead and push a fix to that branch if you like, otherwise i'll handle it when i get to it (maybe tonight) <taylanub>Since we have typed arrays, let's implement ASM.guile!! :P ***vicenteH` is now known as vicenteH
<lloda>wingo: can I amend & push or do a patch on top? <wingo>lloda: either way is fine for me <lloda>hm, I get a git error: unpack-objects abnormal exit (??) I'll leave the fix to you... <mark_weaver>lloda: it sounds like your local git repo has become corrupted. <mark_weaver>or that you have a buggy version of git or something. <mark_weaver>it's certainly not a bug in Guile that's causing that. wingo won't be able to fix that for you. <mark_weaver>I just did "git pull" on master and all's good here. <wingo>mark_weaver: were you able to build on your intel machine? <lloda>I wasn't expecting that. 'I'll leave the fix' means uniform-vector? -> f64vector? <wingo>mark_weaver: not just now, i mean :) any time after your 11 hour thing <wingo>i should change the page size at some point... <mark_weaver>well, I rebuilt master on the intel machine a week or so ago. <wingo>what is the maximum page size on mips again? 64kB or 16 kB? <mark_weaver>wingo: here's the thing that I got stuck on: the obvious patch means that the linker declares that something has to be aligned on a 64kb boundary. <mark_weaver>and then the loader sees that and just calls 'mmap'. <mark_weaver>but 'mmap' doesn't actually guarantee that it will be on a 64kb boundary. <mark_weaver>I tried doing a patch that was more proper, but it crashed. <wingo>could be, there's a bunch of dragons there <mark_weaver>wingo: btw, I finally read that DJB post on entropy attacks. thanks very much for the pointer. very interesting, and good to add to my knowledge. <wingo>you're welcome :) i also found it to be yummy food for thought <mark_weaver>but I've updated gnulib since then, so it would have to be redone obviously. <wingo>i think getpagesize is already imported <wingo>it might not be on the list but it's required by something else <mark_weaver>well, it might be imported indirectly, but not directly. <mark_weaver>if we use it directly, I think it should be imported directly, no? <mark_weaver>anyway, the patch produced a totally broken guile :-( <wingo>i use it in vm.c i think, to figure out how much memory to give back from the stack <wingo>which, btw, seems to be doing well now, no? <wingo>i think i must have had a size_t underflow that ended up returning the entire address space to the kernel :) <mark_weaver>anyway, I have to go afk for a while, and then offline for a while after that. I hope to be online later today. <mark_weaver>(I wouldn't be surprised if that patch shows my ignorance of ELF; I confess I don't yet understand the finer points) <madsy>davexunit: Did you figure out your config path thingie? <davexunit>madsy: do you know anything about how to best store the pkgdatadir in a guile program? <wingo>davexunit: i write it in to a config.scm file usually, at install-time... <mark_weaver>davexunit: configure uses it to generate guix/scripts/guix, substituting in some variables. <wingo>right, that would be a good example <davexunit>I was having trouble finding the prior art. :) <mark_weaver>and then there's some simple code in there to expand out the variable references. <wingo>see config.scm and install-data-local <mark_weaver>hey madsy! I'm going offline very soon for a while, but a little I'd be curious to hear about how your code worked on guile on windows. <wingo>probably there are better ways that use more guile-ish things :) <civodul>dje42: congrats on the recent GDB commits! :-) <civodul>i think that lays the foundation for neat birthday potluck hacks ;-) <madsy>davexunit: Try calling (%package-data-dir%) <davexunit>madsy: I believe that get's the package data dir that guile was built with. <madsy>Can't you put your own library there? <davexunit>that has its own data files in /usr/share or wherever. <add^_>davexunit: you could look at wingo s configfile for guile-figl <madsy>davexunit: What is package-data-dir for if not data? <davexunit>add^_: wingo gave me a few examples. I guess that would be another example. :) <davexunit>(%package-data-dir) yields guile's data dir. I have a datadir for my own application that needs to be baked into my application at make time. <madsy>Why can't your application's data directory be a path which is relative to one of guile's many config paths? <madsy>Sounds like I misunderstood your problem. Ignore me. ***Fuuzetsu is now known as Guest76126
***Guest76126 is now known as Fuuzetsu
***Fuuzetsu is now known as Fuuzetsu`
***Fuuzetsu` is now known as Fuuzetsu
<foeniks>when I import a module in the repl using use-modules, I do not get line and column information with lexing/parsing errors <wingo>try with --auto-compile=fresh and see if there is a difference? <lloda`>it's me, I have erc open at work... <wingo>i was going to ask you to take the patches from here but actually i'm just applying them <lloda`>the next five patches or so go together and are a near complete rewrite of scm_ramapc. <lloda`>I started fixing edge case bugs (empty arrays, etc) and then I thought I could unroll in more cases, etc. <lloda`>there's a fair amount of testing attached so I think it holds up <wingo>yeah i see that, that's great <lloda`>some of the commit messages don't make much sense after all the work on uniform-, etc. <wingo>thanks a lot for this patchset! <wingo>and now you have commit access so hopefully we won't get in this situation again <wingo>(1) please check your commit log format; see ludo's for the best example. there are emacs modes to help, and I'm sure other things as well. <wingo>(2) use complete sentences and capitalization in comments, usually <lloda`>(1) right, I noticed that later :o/ <wingo>(3) interface changes are delicate :) please get buy-in from guile-devel before changing things in .h files, or other publically-visible changes <lloda`>(3) in my defense ;p what I got from the ml wasn't super-constructive <lloda`>so I just started anyway to see where I'd get <wingo>(4) these restrictions notwithstanding, please feel very free to hack on guile :) if what you are committing looks obvious to you, just post to the list, wait a few days, and if there's no comment go ahead and push <wingo>feel free to push tests or bugfixes without such a wait, i think, though sending patches to the list is nice <wingo>and in our defense of (3), there were quite some changes and managing the deprecating story between 2.0 and 2.2 was a bit delicate... but anyway, that part at least is behind us now :) <wingo>it's wonderful to have someone care about the arrays code :) <lloda`>(4) noted and thanks again. I'll try and be careful, as I use this part of Guile heavily mysef. <davexunit>wingo: what emacs modes do you know of for commit logs? <wingo>davexunit: I use magic, pressing C on patch hunks ***altphi_ is now known as altphi
<davexunit>wingo: whoa, didn't know magit had that feature. awesome! <wingo>there are probably better solutions... <wingo>anyway, got to get away from this machine :) thanks again and good night!