IRC channel logs
2023-08-08.log
back to list of logs
<rlb>daviid: sure -- it's mostly just the "three form" case body. One sec... <rlb>Line 255 (new code) there is one of the examples. <rlb>If you have three forms in the case clause, the second is a guard, and the clause matches only if the pattern matches *and* the guard isn't false. <rlb>(That code might also be notably different in the current tree, fwiw -- not sure if that's before or after I switched lokke to use guile 3 exceptions.) <daviid>rlb: ok thanks! so you move lokke to codeberg then, unhappy with sh? just curious <rlb>I currently push it to codeberg, gh, and sourcehut, and while I do have concerns about gh, that's where much/most of the clj community seems to be, so I'll likely maintain the repo there. Though leiningen did fairly notably officially move a bit back: https://codeberg.org/leiningen/leiningen <rlb>People can get it from whichever place they prefer atm. <daviid>ah ok, didn't know you were maintaining all these git repos <rlb>...codeberg's also still not very strong on the ci front, where sourcehut seems pretty good so far, allowing, in particular direct testing of various bsds, etc. as compared to the gh's very limited selection. <rlb>...do the tests not run in parallel? And if they don't, is there any hard constraint there? <rlb>ACTION has been running the full suite a *lot* lately. <rlb>(I'd imagine some tests might not be able to run in parallel, but would hope most could.) <daviid>rlb: tx - sh is the best, after savannah ofc :) - notabug is nice too, but i use savannah ... <rlb>If the tests *can* be parallel, I might look in to it at some point. Would have saved me a chunk of time. <daviid>cool, g-golf now supports user defined signal(s), at last :), better later then never :) - like guile-gnome (though the signal spec in g-golf is slightly different), it does so by processig #:g-signal class definition option(s), fwiw here is a (very simple) example - https://paste.centos.org/view/28af1785 - <rlb>answer - we intentionally disable parallel tests... (see serial-tests in configure.ac) I might investigate later. <rlb>Looks like switching to the parallel test driver (and one check-guile invocation per foo.test) still runs make check in nearly the same amount of time with one core, and goes twice as fast with 3. That'd be nice. (More cores don't currently help right now because (I think) we have a few longer tests.) <rlb>wingo: is there a strong preference against the parallel test driver? If not, I might pursue a bit further later (after the utf-8 stuff). <XiaoBiHu>Hi, I'm trying to find out possible licenses accepted by a module (guix licenses) in guile. Could you please point me out to how do I find out? <rlb>civodul: do you know if there's notable opposition to the automake parallel test driver? I ask because running the tests in parallel takes about half as long here with 3 cores (more don't help atm) and I've been running them a *lot* :) If there's not, I might pursue it a bit further later (after the utf-8 stuff). <rlb>(It would likely change the output behavior, though we can also probably make some adjustments there too if needed, via a custom driver.) <rlb>Anyway it's fairly easy to get the basics working, so I could demonstrate it later, though didn't want to bother if we were already sure we weren't interested.