<brendyyn>I keeg getting TLS error in procedure 'fill_session_record_port_input': Resource temporarily unavailable while downloading 4g18cdg...gcc-5.4.0.tar.xz from hygra.gnu.org, although it seems to happen at a different point each time.
<paroneayea>not 100% sure the cpu frequency scaling daemon is doing anything (it doesn't know what my CPU is)
<reepca>I think it might somehow be related to the autocomplete (or rather, auto-suggestions-popping-up) functionality - I started an emacs with GUILE_LOAD_PATH and GUILE_LOAD_COMPILED_PATH set to empty strings, and it's running smoothly
<apteryx[m]>Would make sense, as I saw in Geiser manual that there is a limitation where if you edit aother Guile buffer and while running commands at Geiser the Geiser buffer won't be able to refresh.
<brendyyn>Well if gnu/packages/version-control.scm is a directory ~/foo, and you have GUIX_PACKAGE_PATCH=~/foo set, then it will automatically be available to guix and you can just run guix package -i bar
<solene>amz3`: I'm a bit lost, I find guix really disturbing, my first though was "another one linux distro doing something a bit different" but in fact, after installed it, I find it's a totally new system of its kind using linux as a kernel, not another one distro replacing X by Y. I like it
<ng0>is iyzsong around? I wonder about the difference of how git-daemon-service ended up and let's say bitlbee-daemon, the part which defines the commands. it's difficult to debug (darkhttpd)
<ng0>my second idea is that darkhttpd is just awful.
<amz3`>solene: that's what is called (disruptive) innovation
<DoublePlusGood23>bavier: I can get swish-e to compile and pass the tests, however it requires disabling both `zlib` and `libxml2`, unsure how that effects how people are using it, they seem to be ""just"" performance improvements, but please take a look at my commit to guix-patches.
<OriansJ>janneke: I unfortunately had a less productive week, (5 commits) due to the fact I had to convert the stage0 lisp from dynamic scope to lexical scope and then I could finally add let expressions. (also main repo for stage0 is no longer github it is https://savannah.nongnu.org/projects/stage0/ )
<janneke>Hmm, that sounds like a great transition though!
<janneke>How did you add let? I added macros and built everything on top of that.
<DoublePlusGood23>How do I switch to guile 2.2? reconfiguring my system didn't seem to fix it?
<janneke>DoublePlusGood23: what do you mean by `switch'
<DoublePlusGood23>janneke: `ERROR: in procedure load-thunk-from-memory: no such file or directory`
<OriansJ>janneke: the struct cell* process_let part down is the only bit you actually need
<janneke>OriansJ: ah, great. I've been struggling with pushing all posible non-primitives out, maybe too much so.
<janneke>OriansJ: I'm wondering [again :-)] how to continue, next up seems new features like: floats, math funcs, posix stuff (fopen, fwrite, fork, exec)
<janneke>Not sure how to determine which are really needed, all of them?
<OriansJ>janneke: I'd argue only fopen, fclose, fputc (possibly fprintf if you want)
<OriansJ>janneke: since compilers don't actually need floating point, math funcs or anything fancy
<janneke>OriansJ: I was hoping that...but tinycc's test suite mentions all of those
<OriansJ>janneke: that is because tinycc has to support all of them to be C99 compliant but we only need to support the subset required to build tinycc
<janneke>mescc has the unixy open, write, read etc...i would only need to mimick the microsoft-friendly(?) `f'-variants
<janneke>OriansJ: rrrright! So I should now be switching to attempting to build tinycc, possibly.
<OriansJ>janneke: well once you can build tinycc, we can celebrate as we will have completed a major milestone
<ng0>does someone have a strong opinion on darkhttpd running without dropped root privileges when it's inside a guix container? the default is not to drop, and it has problems accepting the users+group
<lfam>Anyways, that's how it worked for me, but I didn't know what to expect exactly, so my observations were vague
<janneke>OriansJ: i wanted the mes C core to be the absolute minimum, thinking we would need to hand-code in in hex or assembly. A binary program reader that loads the .mo file is much smaller than the minimal scheme/sexp reader. Possibly that's over-optimization and it also introduces another binary...oh well :-D
<OriansJ>but you are absolutely correct, yours could be implemented much easier in hex than mine
<ng0>for containers, is (source "/some/dir/or/device") static or can I use (source hashdollarhtdocs) (my irc client interprets this, sorry) aswell when this is defined?
<janneke>OriansJ: I was probably overreacting with my reader blob. At the time I also hoped to produce dumps from bigger programs (whole of mescc) and gain performance. It proved not easily feasible to do so and also performance wise it doesn't really help. Possibly time to remove that feature or at least not depend on it and add the sexp reader in C :-)
<lfam>andreh: In 'Running Guix Before It Is Installed', the script `pre-inst-env` is mentioned. Prefix your Guix commands with that script to use the copy of Guix in your Git tree
<OriansJ>also, I'll be extending stage0 lisp with any primitives you want janneke
<lfam>Finally, I don't recommend using `guix package --install` to test packages. Instead, use `guix environment --ad-hoc`. Otherwise, you'll end up with all your work-in-progress copies of the packages protected against garbage collection, since they will be in your profiles
<lfam>Of course that's fine if you want it, but typically one does not
<lfam>civodul: I'm confused about `make release`. The for-loop that builds the disk images doesn't seem to set up offloading or target any system type specifically.
<andreh>lfam: Thanks for the tip. I'll try using `pre-inst-env`. I overlooked sections 8.1 and 8.2.
<ng0>grooarr... I somehow created a default field for service which is ().. i thought an empty list is okay
<lfam>paroneayea: Oh yeah, about thermald! Your machine overheats when you push it too hard?
<ng0>could it be that darkhttpd doesn't understand how to drop privileges for some stupid reason when run as a system service? when I exectute it as my user it works. I used to be able to run it as root. once user+group gets in I get error messages:
<ng0>ERROR: in procedure getgrp: No such file or directory
<janneke>OriansJ: I have no simpler plan, and the plan looks sound. Possibly quite some work, but if we first manage to close the bootstrap loop with the mes.c hack, we can then work steadily on going really full-source using stage0.
<civodul>i was concerned that this might be a forgotten branch
<lfam>civodul: `make release` fails for me due to a TEXI2DVI error: "TEXI2DVI doc/guix.dvi [newline] /gnu/store/kqpa34yv6dfbkfm5zryr07rph2nfs0a5-profile/bin/texi2dvi: tex exited with bad status, quitting."
<civodul>lfam: yeah you need to copy build-aux/texinfo.tex from a recent Gnulib copy