<rlb>...hmm, lately, I've wanted a "clobber the cache for these paths" command. We don't already have anything like that do we? e.g. "guile --clear-auto-compiled --recursive ." or something. <rlb>I want to be able to run that when hacking on guile itself, and/or some other project, in conjunction with a clean, to avoid having to say unpleasant things when "macros"... <rlb>Right now I just "rm -rf ~/.cache/guile/ccache/.../the/relevant/path", but that's less than ideal. <str1ngs>rlb: that sounds useful. I selectively remove ~/.cache/guile myself. or selectively use --no-auto-compile <ArneBab_>leoprikler: whenever a good video is playing I just run likeplaying from anywhere and it grabs the filename from the mpv window :-) <rlb>str1ngs: right, and no-auto-compile doesn't work very well when I'm running a "make -j 7 check" with many more than 7 independent tests :) <rlb>it's very expensive, though I'm pretty sure it did work, which surprised me a little, i.e. good sign that we're doing the right thing with respect to concurrent auto-compiles from multiple processes. <rlb>fsvo "right thing", since the semantics there could be interesting without some care... <rlb>even if you make everything atomic. *rlb would have to think about it -- maybe it's fine. ***slyfox_ is now known as slyfox
<daviid>I still haven't looked into this, not even superficially, I will of course, asap, but I thought you might be iterested to read (or re-read) this ***jonsger1 is now known as jonsger
<ArneBab_>I just restructured the guile template in conf: conf new -l guile PROJ now creates PROJ/PROJ.scm along with the module (PROJ PROJ). In short: conf new -l guile foo ; cd foo ; make install ; foo ***catonano_ is now known as catonano
<daviid>it points to the NEWS entry from andy, and proposed a manual solution to the 'problem' - which is what str1ngs rightfully did in his proposed patch - however, in g-golf (and guile-cv ...) I need an automated solution to this, and, hopefully, backward compatible ... also, <daviid>the re-export of the public interface of a module, which for some name(s) replaces guile core binding(s), is one of the two problem g-golf needs to solbvve (automatically, by running a program, not by manually writing in a file) <daviid>leoprikler: the results/solution must be so there are no warnings, since the replacement is deliberate ... fwiw <daviid>what ever it takes, it must work jut like it does for 2.2.7 - in tha paste used by dsmith-work to post to the ml <johnjay>hey. i'm trying to compile something based with guile but i can't get past the configure script <johnjay>i got rid of guile 3.0 as the guy said but the script says it can't find guile lib <johnjay>i have guile 2.2 and 2.2-dev and libs. but it says configure: error: guile-lib is missing; please install it. <johnjay>ah ok. in u buntu you need "guile-library" package. that solve it <catonano>johnjay: may I ask you what are you trying to compile ? <johnjay>but if you have something more common i can try to compile let me know <johnjay>also what's teh deal with guile3 and guile2 existing at same time? is there big incompatibility? <johnjay>i guess ubuntu 20 didn't install that library by default <leoprikler>guile3 and guile2 are entirely separate, they shouldn't even look into the same module paths <leoprikler>daviid: Making it work "just like Guile 2.2.7" is merely a matter of whether or not you add (hashq-set! (module-replacements iface) key #t) after (module-add! iface key value) <leoprikler>btw. is it wanted that all method-gf are pushed back into hl-api gobject? <nerdypepper>i am looking to have some sorta integration with guile-lint <nerdypepper>also, guile-lint seems to dislike using square brackets [] in let and cond expressions, any way to allow that? <str1ngs>leoprikler: ah I got it sorted out in #emacs re emacsy was the package. <leoprikler>I have no idea what question that was, you'd have to help my memory out here a bit :) <str1ngs>it was some time ago. reguards to guile 3 and unbuntu <str1ngs>the project he was building was emacsy. which should support guile 3 soon. I have a local working copy anyways. <leoprikler>once emacsy is ported to guile 3 nomad should be no problem, right? <str1ngs>actually I did'nt know ubuntu had guile 3 in the repo. I've been building in manually <str1ngs>the only blocker is g-golf guile 3 issue which is no rush. once that is done g-golf looks pretty good for guile 3 as well. <leoprikler>I don't know anything about newer ubuntu period. IIRC "stable" is still stuck on 18.04 <leoprikler>The only Debian system i still have is a LM installation I plan to nuke after getting a grade on my masters thesis. <str1ngs>which has much better guile package support than 18.04 ie guile-ssh guile-git etc <leoprikler>Well, that certainly makes it easier to bootstrap Guix <str1ngs>yes, I forget who did the work. but it tricked down to ubuntu nicely. thank unsong hero! <str1ngs>johnjay: I did find the reason for the failure out of tree builds and emacsy. it's related to the not enabling all of the examples. A regression introduced by adding that --without-examples flag. thanks for finding that, I should have that fixed soon. ***nikita` is now known as ronja`
***ronja` is now known as nikita`
<apteryx>is there anything to manipulate plists in Guile? <apteryx>plist being of the for '(key1 value1 key2 value2 ...) <leoprikler>for fixed combine let-keywords with quasiquote, for arbitrary no <leoprikler>(let-keywords plist #f ((key1 default) (key2 default) ...) `(#:key1 ,new-key1 #:key2 ,new-key2 ...) ) <apteryx>are the ellipses part of the supported syntax? <apteryx>I guess not. The keys I want to manipulate are from an expected set, but the number of keys can vary, so I think let-keywords won't help for that. <leoprikler>you can set the other keys to a well-known default e.g. #f or *unspecified* <apteryx>ah right, if I specify the full set it can work, but that's on the verbose side and difficult to maintain <leoprikler>pretty sure you could write a macro for that, but I see your point <apteryx>nothing standard to get all keys of an alist, right? I need to make my own match-lambda <pkill9>actually that's what I want to do <brettgilio>wingo: hey, I can't parse your ATOM feed. I think the formatting on it might be broken? <str1ngs>dsmith-work: that change looks promising. does that help with guile 3 map issue? and work for guile 2 as well? <dsmith-work>str1ngs: Pretty sure that won't work with guile 2. I don't have a 2 install handy. Should be possible to work with both by conditionally calling the hash-set! (I think) <dsmith-work>leoprikler: Yeah, the way replacements are recorded changed. <str1ngs>dsmith-work: gotcha, thanks for looking at this BTW. <dsmith-work>from the NEWS file regarding #:re-export-and-replace: "Note to make this change, we had to change the way replacement flags are <str1ngs>is it possible to get major version number someway? or do I need to use (version) <apteryx>It's a bit more complicated than I'd like, but it does exactly what I want. <dsmith-work>str1ngs: Never tried. But (apropos "version") returns quite a few results.. <str1ngs>ahh (major-version) what I need thanks. I just need it for ~/.guile. ***evhan` is now known as evhan