<drakonis>cwebber: i figure you'll might be able to best answer this, fibers or 8sync and will 8sync's fibers port ever be completed? ***sneek_ is now known as sneek
<cwebber>fibers, or goblins, which runs on top of fibers too, and will have an 8sync compatibility layer soon... there will be a whole presentation showing it off at fosdem this year, and a release during this next week :) :) <drakonis>i'll definitely it for shepherd and guix then <drakonis>guess its time to play with goblins then! <cwebber>drakonis: I need to finish the hall bindings <cwebber>and then I need to make a guix package <cwebber>goals for the next couple of days :O <cwebber>g'nite drakonis, hope you're well :) <drakonis>time to be the change i want to see in the world :v <drakonis>i should figure out how to fix the load path while on guix <drakonis>if i use guix on a foreign distro, it works fine <drakonis>also relevant: looks like fibers has seen changes not covered in 1.0.0 <cwebber>drakonis: I just had an email exchange about that today ;) <drakonis>that said, the readme no longer refers to the concurrency as erlang style, but as golang style <cwebber>which is good because it includes an important memory leak bug I helped with which will be key to guile :) <cwebber>well, it's kind of true, it's more along the lines of CSP than actors, but *goblins* layers actors on top of CSP, so actors come back :) <drakonis>with that said, i have plans for goblins. <drakonis>but don't let me stop you from going to bed. <cwebber>drakonis: the racket docs mostly apply ***spacebat4 is now known as spacebat
<unmatched-paren>how can i configure repl options in .guile? repl-default-option-set! doesn't work (unbound variable...) <tohoyn>unmatched-paren: try to add (use-modules (system repl common)) into before repl-default-option-set! <stis>Did fix the python on guile implementation so that it compiles again inhead. <stis>also works with latest guile <tohoyn>daviid: have you looked at the bug with <g-settings-action>? <daviid>tohoyn: no, i fixed the <g-simple-action> [g-variant related] bug, and to do that, while rewriting your example, i dropped that part ... <tohoyn>daviid: just a thought: could the g-golf API take types as GOOPS classes instead of symbols? For example in (make <closure> ...) the #:return-type and #:param-types arguments. <rlb>civodul: added the lokke ./configure check for the broken 3.0.7 versions, fwiw. <civodul>i suspect psyntax adds too much cruft to the source property table <civodul>it adds syntax properties to lists of syntax objects, each of which already has location info <rlb>civodul: not yet, but might be a path issue if you're building differently for guix. Alternately, I think I've seen stale bits in ~/.cache/guile/ccache cause trouble. Might try make clean, and clobbering that too. <rlb>Hmm, should perhaps also add a "make distcheck" to ci... <rlb>(trying another distcheck here) <rlb>Are you doing anything interesting (wrt configure args, etc.)? <civodul>rlb: nothing particular, and it's in a clean environment without ~/.cache <rlb>And is this against a mostly "plain" 3.0.7, building from a lokke "make dist" archive? <rlb>The ci builds are working, but I should probably change them to run from a dist archive. I'll try a manual unpack/build in case distcheck is cheating somehow... <civodul>this is Guile 3.0.7 + d79a226359d28f4a1dc5df136e5544d699903a96 <rlb>ok, I suppose both in ci (ubuntu) and locally, I'm using debian's version of guile, which does have some other patches (though don't offhand recall if they're notable). <rlb>I guess if you haven't might also try a clean tree, i.e. `git clean -fdx` or a new clone or... <rlb>(particularly if that tree's been used across the move to drop git as a build requirement) <rlb>ok, "make -j5 check" on current main worked here from a clean (git) tree. *rlb looks at the other debian patches... <rlb>clojure.core is a little tricky -- offhand, I wonder if the path lookup is failing for "reasons", though no idea why that'd be different for you. <rlb>I can try building a local 3.0.7 + syntax-fix in a while and see if that makes any difference. <civodul>oh wait, could it be that compiled-file-name return #f because HOME is unset? <rlb>hmm, wonder if it's a path problem. <rlb>i.e. that code assumes some things that might not be ok -- will need to delve... <rlb>And yeah, won't be at all susprised if the current version somehow depends on things like HOME... <rlb>Some of the stuff in ns.scm will almost certainly need improvement. Various bits I'm already not sure about. <civodul>so i seem to have a Guix package \o/ <rlb>nice, and thanks for helping discover the problems :) <rlb>I'll put "mess with HOME" on the list... <daviid>sneek: later tell tohoyn I don't want todo that, for various reasons, but of course feel free to do it in theme-d[-golf] if that suites your needs 'better'