IRC channel logs

2020-02-03.log

back to list of logs

***jao is now known as Guest64080
***Guest64080 is now known as jao`
***jao` is now known as jao
<fnstudio>Hi, if I were looking for something comparable to Python f-strings, where should I start from?
<fnstudio>or something comparable to Python's `.format()`
<fnstudio>i.e. a way of injecting variables in a string
<lloda>fnstudio: (ice-9 format)
<fnstudio>lloda: super, thanks!
<lloda>yw
<dsmith-work>{appropriate time} Greetings, Guilers
<erkin>Heya
<dsmith-work>sneek: paste?
<sneek>Last time I checked paste is https://paste.debian.net
***ng0_ is now known as ng0
<rlb>guile-3.0 has been approved for debian unstable/sid: https://buildd.debian.org/status/package.php?p=guile-3.0&suite=sid Looks like we have some work to do on armel, mipsel, and s390x, though with luck, maybe mipsel just needs a longer timeout...
<rlb>And of course it'd be nice if we can track down the failures on the non-release archtectures too...
<rlb>Do we expect the build time on non-jit platforms to be about the same as 2.2?
<dsmith-work>Woo!
<jonsger>rlb: I think I have the same build failure on armv7l on openSUSE like on debians armel https://build.opensuse.org/package/live_build_log/devel:languages:misc/guile/openSUSE_Factory/armv7l
<jonsger>rlb: I think the build time on non-jit platforms is longer then on 2.2. That's at least my impression
<jonsger>rlb: I have data: build time on non-jit platforms is around factor 2x. e.g. https://build.opensuse.org/packages/guile/job_history/devel:languages:misc/openSUSE_Factory/ppc64
<rlb>Yeah, was asking because it looks like some may have been coming in a 7h vs 3h for mips64el, but there may also be buildd host differences...
<rlb>ahh, ok, thanks.
<rlb>Looks like you saw an illegal instruction where the deb saw a segmentation fault?
<jonsger>yes
<jonsger>our armv7l build is in qemu of aarch64 builders. Is "hasse" a native armel build node?
<rlb>looks like real hardware: https://db.debian.org/machines.cgi
<rlb> https://db.debian.org/machines.cgi?host=hasse
<jonsger>maybe I just should setup the little armv7 board I have on my desk :P
<rlb>Looked like s390x might not have even made it through configure -- maybe I'll poke at that one first. Had trouble in the past with s390 (though I think that was emacs) because it was a 31-bit arch, but s390x shouldn't have that issue iirc.
<jonsger>rlb: s390x is fine on openSUSE: https://build.opensuse.org/package/live_build_log/devel:languages:misc/guile/openSUSE_Factory/s390x
<rlb>ok, thanks. Sometimes it's also the default debian build options -- we've had to omit some in the past.
<rlb>e.g. maybe stack-protector?
<jonsger>rlb: it gets build with those flags https://paste.opensuse.org/view/raw/84512338 plus the ones in the spec file
<rlb>jonsger: ok, thanks again.
<jonsger>without HDMI cable, SD card and u-boot image for the board it seems to be pretty hard
<rlb>hah
<drakonis>what's the current guile package manager?
<drakonis>is it still guildhall or guix replaced it?
<jonsger>drakonis: I think guildhall is not really used anymore, so kind of replaced by guix :)
<civodul>rlb: great that guile-3.0 was approved for Debian!
<civodul>there's a couple of bugs to address on ARMv7 and AArch64, but yeah, we'll get there
<stis>please also the nasty bug#39315
<dsmith-work>Any other Guile talks at FOSDEM?
<civodul>dsmith-work: there were talks about things around Guile, but only one on Guile proper i guess
<civodul>see https://fosdem.org/2020/schedule/track/minimalistic_experimental_and_emerging_languages/
<civodul>and https://guix.gnu.org/blog/2020/meet-guix-at-fosdem-2020/
<dsmith-work>thanks!
<dsmith-work>Ooo! Guix instead of yocto!
<civodul>i find it nice to see all these things being done with Guile!
<civodul>heh
<civodul>that one was great IMO
<civodul>in a way we're all shamelessly taking advantage of Guile goodness :-)
<dsmith-work>I have *never* gotten anything building, let alone working, with yocto.
*stis use buildroot
<dsmith-work>Buildroot is my tool of choice
<jonsger>rlb: did you made progress on armel?
<rlb>jonsger: no - haven't had time to mess with it yet, and may not for a bit.
<rlb>i.e. might be "days".
<jonsger>oke, then I'll just open a bug. As I'm struggling hard to get an armv7l qemu guest running on aarch64...
<rlb>when I get around to it, I can test directly on a porterbox, which may or may not help me :)
<sneek>.oO( Hours seem like days... )
<drakonis>how did guix days go?
<rekado>I’m working on the Guix Workflow Language again and the compilation of my modules takes a *very* long time.
<rekado>They are not complicated modules at all.
<rekado>long time = several minutes
<rekado>this is with Guile 2.2.6
<rekado>The modules use GOOPS, so I think that might be why; but is there a way to profile the compiler and figure out what is taking so long?
<rekado>the test suite also runs very slowly even though the tests are rather simple.
<civodul>heya rekado!
<civodul>rekado: you could try: ,pr (compile-file "...")
<civodul>or perhaps ,time to start with (to see if it's a GC issue)
<rekado>thanks, I’ll give it a try
<rekado>huh, it takes mere seconds when comping like that.
<rekado>oh…
<civodul>oh, could it be that you're evaluating many of your dependencies?
<rekado>YES!
<civodul>e.g., because GUILE_LOAD_COMPILED_PATH isn't correct
<rekado>all of Guix
<civodul>ah! that takes a while :-)
<rekado>embarrassing…
<civodul>heh, that happens
<rekado>the “env” script was wrong
<civodul>we'll pretend nothing happened
<civodul>hey rekado, hi!
<rekado>oh hello!
<rekado>long time no see
<civodul>:-)
<civodul>we missed you in Brussels!
<rekado>I really wanted to attend. I missed the energy, the collaboration, and the joy of meeting fellow Guix hackers.
<rekado>I’m looking forward to reports from the Guix Days and FOSDEM
<civodul>yes, we'll hopefully send notes soon
<MutoShack>I'm having trouble with the SXML modules
<MutoShack>Is there a way to output "<!-- -->" HTML comments from sxml->xml ?
<RhodiumToad>apparently not
<RhodiumToad>but,
<RhodiumToad>you can put a procedure in the sxml tree, which will be called with output directed to the appropriate port
<RhodiumToad>(and obviously that can write anything it pleases)
<MutoShack>Thanks, RhodiumToad. I'm in no hurry, so I'll just put it in my todo list for now!
<RhodiumToad>sxml is a bit hemipygian in places