IRC channel logs

2020-01-14.log

back to list of logs

***rlb` is now known as rlb
<jcowan>make check on Cygwin has 2 minor errors, probably newlib/glibc incompatibilities in strftime
<daviid>jcowan: but iirc, you pass -1 a the jit option, which means iirc, no jit, that's not good
<daviid>this means that 3.0 perfs will be inferrior, o cygwin, to 2.2, in general
<daviid>jcowan: or was the jit option just for the make step? not sure, worth asking i guess ...
<daviid>str1ngs: i pushed a series of pactahes to document the <gdk-event>, <gdk-event-key> class, procedures, accessors and methods
<daviid>also, the gdk-event-key:state now returns a proper list of state modifiers, like, for example, '( shift-mask control-mask mod5-mask), not just the state bitmask intenger
<daviid>*patches
<daviid>the state modifiers doc entry points to the GdkModifierTyp doc, so one can read more about this s well ...
<daviid>have to go, bbl
<mwette>wingo: wrt #lang, a while ago I posted some code for #lang to bug 30094; and it's also under github.com/mwette/guile-contrib.
<str1ngs>sneek, later tell daviid thank you. Will take a look first thing tomrrow.
<sneek>Got it.
<sp1ff>Hi-- I have a question around Autotools & Guile. I've found examples of how to write Makefile.am for Guile modules, specifically around compiling the Scheme and installing the .go files. I'm trying to write a Makefile.am for a Guile script; that is, a file that starts with "#!/usr/local/bin/guile -e main -s..." and gets installed into ${prefix}/bin.
<sp1ff>What are Guile's rules for where the .go files should go for scripts?
<sp1ff>Just ran strace on a test script & I see guile looking in the usual place: /usr/local/lib/guile/2.2/ccache//usr/local/bin/test-script.go
<sp1ff>I'm still wondering if this is documented anywhere, and whether anyone else has done this?
***apteryx_ is now known as apteryx
<jcowan>sneek, later tell daviid yes, only during make
<sneek>Got it.
<mwette>sp1ff: I think GUILE_SITE_DIR config macro will set GUILE_SITE_CCACHE which is where you should install .go files; check guile.m4
<mwette>wingo: if you are looking for future efforts please consider working on the debugger; debugging in 1.X was much more satisfying than it is in 2.X
<mwette>wingo: I tried some hacks but fell short: for example, added a "-g" flag to "guile compile" and then tried to use that to remove reuse of allocated slots
<sp1ff>Thanks mwette
<dsmith>Hey hey guilers. Started building 2.9.9 on my beaglebone. Getting that socklen_t problm from configure.
<dsmith>Looks like I'm running wheezy. I fear I'll neet to upgrade first.
<dsmith>checking for socklen_t equivalent... configure: error: Cannot find a type to use in place of socklen_t
<dsmith>
<drakonis>oh man, wheezy is old
<drakonis>its nearly 7 years since it released, its definitely time to go up a few versions
<dsmith>Yeah. Question is, should I just jump to current stable, or walk through each release.
<dsmith>So next would be Jessie
<dsmith>Here we go. Upgrading to jessie...
<drakonis>actually jessie is also EOLed
<drakonis>after jessie you still have to go through stretch to reach buster
<dsmith>Yes
<dsmith> https://ewen.mcneill.gen.nz/blog/entry/2017-06-07-debian-wheezy-to-stretch-in-two-larger-steps/
***sneek_ is now known as sneek
<dsmith>sneek: botsnack
<sneek>:)
<dsmith>~uname
<dsmith>~uptime
<dsmith>sneek: uptime
<dsmith>Hmmm.
<dsmith>sneek: uname
<dsmith>sneek: seen rlb
<sneek>rlb was here Jan 13 at 02:11 am UTC, saying: wingo: thanks wrt fluid-ref -- obviously I hadn't attempted it yet..
<dsmith>!uptime
<sneek> 00:19:33 up 7 min, 2 users, load average: 1.09, 1.12, 0.51
<dsmith>!uname
<sneek>Linux beaglebone 3.8.13-bone47 #1 SMP Fri Apr 11 01:36:09 UTC 2014 armv7l GNU/Linux
<dsmith>sneek: seen wingo
<sneek>wingo was here Jan 13 at 08:41 pm UTC, saying: jcowan: does it work if you compile with GUILE_JIT_THRESHOLD=-1 set in the environment?.
<dsmith>sneek: botsnack
<sneek>:)
<dsmith>yey, running jessie
<dsmith>!uname
<sneek>Linux beaglebone 3.8.13-bone47 #1 SMP Fri Apr 11 01:36:09 UTC 2014 armv7l GNU/Linux
<dsmith>sneek: info
<dsmith>sneek: version
<sneek>Sneeky bot running on Guile version 2.0.11 using bobot++ 2.3.0-darcs
<str1ngs>sp1ff: if you are writing a shebang script with autotools I would not compile it . And I would use configure.ac to substitute the guile path. this makes your script more portable.
<str1ngs>sp1ff: typically these files are have an .in extention. you can also use Makefile.am and sed to do the substitution
<str1ngs>as well
<dsmith>Bah! Running out of "disk". Currently using 1.4G of 1.7G.
<dsmith>180M free
<dsmith>Well, guile ./configure'ed on jessie. Doesn't like the libreadline though.
<dsmith>Let's see if it finished before tomorrow morning...
***ng0_ is now known as ng0
<wingo>hoo, doing a benchmark run takes around 2 hours :P
<civodul>wingo: for two different Guiles?
<civodul>that's why you should make Guile faster ;-)
<wingo>for one guile
*jonsger prepared Guile 3.0 on openSUSE Tumbleweed :P
<wingo>i just made a change so that (eq? proc proc) would be true, but it means i need to re-run benchmarks
<civodul>oh, i see
<wingo>but no need to re-run old benchmarks to make a comparison
<civodul>i've prepared for package migration to Guile 3.0 in Guix
<civodul>good news is we have quite a few Guile packages these days :-)
<civodul>and many run just fine on Guile 3
<wingo>what kinds of problems do you see? :)
<jonsger>civodul: I saw it. I proposed patches for guile 3.0 to https://gitlab.com/mothacehe/guile-parted/merge_requests/1 and https://gitlab.com/mothacehe/guile-newt/merge_requests/1
<civodul>wingo: that's the thing: i don't see problems so far!
<civodul>apart from adding "3.0" in configure.ac, like jonsger writes
<jonsger>and updating m4 for guile 2.9.x
<civodul>yes, "guile.m4"
<civodul>i still to come back to Guix itself though, there were some issues, but hopefully nothing terrible
<wingo>jonsger: great!
<wingo> https://wingolog.org/pub/guile-2.9.7-vs-guile-2.9.9-with-eq-change-microbenchmarks.png
<wingo>looks like the eq change is acceptable
<dsmith>Aww. Fouey. Compile failed.
<dsmith>wingo: Can 2.9.9 be compiled for 32bit arm?
<dsmith>I might need a newer compiler
<dsmith>I have 4.9.2
<wingo>dsmith pretty sure we had a positive report
<dsmith>Hmm. I don't have enough free flash space to upgrade.
<dsmith>wingo: Two kinds of errors in jit.c:
<dsmith>error: initializer element is not constant
<dsmith>warning: left shift count >= width of type
<dsmith>Well, that's a warning.
<wingo>line number for the left-shift thing?
<wingo>or is that in hash.c
<dsmith>Yes, hash.c. 3 of them. hash.c:123:3 hash.c:132:3 hash.c:221:5
<jonsger>wingo: I still have illegal instruction for armv7l https://build.opensuse.org/package/live_build_log/home:jbrielmaier:guile-3.0/guile3.0/openSUSE_Factory_ARM/armv7l but I think this is due to the wrong subset of ARM
*wingo nod
<wingo>realistically guile 3.0.0 is going to have some build and codegen issues that will have to be fixed during the stable series
<wingo>i am ok with that, as i don't have unlimited resources :P
<civodul>yes, i assume there'll be a 3.0.1 someday :-)
<chrislck>mwette: i found out how to use a fender... use the (=> continue) clause as described in http://ceaude.twoticketsplease.de/articles/an-introduction-to-lispy-pattern-matching.html and is not described in the manual at all
<chrislck>so, not obvious at all, but at least it exists
<dsmith>wingo: Fouey, I'm outof time. I'll get you a proper report on the lists.
<dsmith>I might need to do a fresh install on this box.
<mwette>chrislck: thanks!
<chrislck>plus, it seems that we could use the (? predicate?) somewhere deep *inside* the pattern, although I can't seem to find the right incantations yet :/
<jonsger>wingo: the aarch64 build is fine. That's good, because most modern ARM platforms are 64bit :P
<wingo>:)
<wingo>that's good to hear!
<wingo>i did check guile on armv7 and aarch64 but it was a few months ago
<wingo>so good that it's still working
<jonsger>oke, I didn't tested on runtime yet, but lets see
<wingo>if it builds it's probably ok
<wingo>because the compiler exercises itself
<d4ryus>hi #guile, does anyone know of any code that makes use of the GNU Make guile bindings (e.g. to generate templates)?
<OrangeShark>hello everyone
***ng0_ is now known as ng0
<civodul>d4ryus: no, i suspect it's not much used
<rlb>OK, for any interested, timeout has been lengthened drastically for the debian arm* buildds, and 2.2.6 has finally built there: https://buildd.debian.org/status/package.php?p=guile-2.2&suite=sid Should now be able to migrate to testing/bullseye.
<rlb>(well, assuming armel finishes successfully -- armhf was fine)
<chrislck>mwette: using (? predicate? var) in match: \o/ https://pastebin.com/raw/nuTqQPFt
<zig>o/
<zig>I am not sure if the Guile announcement message was changed recently but it really sound good.
<zig>that : https://paste.gnome.org/pubrydvun
<davidl>zig: it is good :-D
<davidl>I recommended guile here: https://www.slant.co/topics/15491/~general-purpose-programming-languages
<davidl>In case anyone's interested; I wrote some examples for using guile-bash and python-on-guile and to extend guile in C, here: https://gitlab.com/methuselah-0/bash-c-guile-python-examples
***drakonis1 is now known as drakonis
<zig>good idea about slant.co, I submitted my review.
<zig>davidl: I did not know it was possible
<davidl>zig; nice. What did you now know was possible?
<davidl>not*
<zig>davidl: I did not know about guile-bash actually
<davidl>If I remember correctly, it was made by some russian math guy whom I think went to prison for a Tor node, but I dont remember exactly. He announced it on Guix a few years ago, but then announced he wouldn't be able to maintain it. It needs updated for Guile 3.0. I managed to fix it for 2.2, but I don't understand the code.
<davidl>it=the guix package for it.
<rgherdt>davidl: nice, thanks for the initiative. Also added some points