<jxself>Well, after thinking about it, perhaps I will take a stab at a modified Firefox. Since this is a whole new thing for me, no promises though. <jxself>tar xf firefox-31.1.1esr.source.tar.bz2 <jxself>I'm not sure where I got 8GB of RAM as being needed to build Firefox. The Mozilla documentations says they recommend 4 and you may have problems with less an 2. <mark_weaver>a while ago I remember reading of problems linking firefox on 32-bit systems, but that have only been on windows and maybe resolved now somehow. ***Basstard1 is now known as Basstard`
***nully_ is now known as nully
<civodul>alezost: if i enable pretty-sha-path-mode in a buffer and then disable it, re-enabling it has no effect <jxself>It was finally idle so I took advantage of the opportunity and rebooted wildebeest to finish the upgrade. <jxself>Also applied the BASH package uprade that was available. <mthl>i'm in trouble right now <mthl>with Guix of cours, because Free Software doesn't work <mthl>it seems like the bug that someone report on guix-devel <davexunit>guile-json is intended to be an optional dependency. <mthl>i've tried to ./pre-inst-env guix pull <civodul>does "./pre-inst-env guix build hello" or similar works? <civodul>(BTW, please use paste.lisp.org for instance, because pastebin.com is inaccessible via Tor) <civodul>so something's deeply broken here :-) <civodul>are there any changes in the source tree? <civodul>is gnu/packages/bootstrap/x86_64-linux/guile-2.0.9.tar.xz present? <alezost>civodul: Most likely it's because `pretty-sha-path-flush-function' variable is nil. Emacs 24.3 doesn't have `font-lock-flush' function (Emacs 24.4 has it), so that var is set to nil for the earlier versions. <alezost>civodul: Could you try again with (setq pretty-sha-path-flush-function 'jit-lock-refontify) <mthl>maybe it has something to do with the fact i remove quite a lot of packages from apt? <civodul>alezost: that does the trick, indeed <civodul>mthl: if the guile-2.0.9.tar.xz file is missing, that cannot work <civodul>normally, it gets downloaded when typing "make" in the source tree <alezost>civodul: ok, thanks for confirming, I'll change the default value for the earlier versions then <davexunit>civodul: thanks for pointing out the weirdness with ruby-i18n. I'll have to do some thinking about how to track references in ruby packages. <davexunit>civodul: looks like Nix removes the .gem file. seems there's no reason to keep it around. <davexunit>I will also need to introduce a wrap-program step <mthl>civodul: configure: error: GNU libgcrypt does not appear to be usable; see `--with-libgcrypt-prefix' and `README'. <davexunit>ah, and they just use the .gem files from rubygems.org... time to rewrite the build system! <civodul>davexunit: ISTR something similar about .egg in Python, which are compressed archives <civodul>mthl: if you installed libgcrypt with Guix, use --with-libgcrypt-prefix=~/.guix-profile <mthl>libgcrypt was install with guix, so i thought sompething was wrong with that version, so i remove to use the debian version but nothing change <civodul>it's fine to use the Guix version, that's what i do <civodul>you just need to tell ./configure where to find it <davexunit>civodul: thanks. I think I have a better understanding of the ruby build system now, thanks to the Nix folks. <civodul>mthl: see "Installing Guix from Guix" in README :-) <civodul>davexunit: Nixpkgs is a useful reference ;-) <davexunit>yeah, I wish I hadn't overlooked this source file before. I got confused because they auto-generate a bunch of the ruby library packages and I wasn't sure what was going on. <davexunit>civodul: not guix-related, but do you have a workflow for running srfi-64 tests from the REPL? <mthl>civodul: ok i'm checking <civodul>davexunit: nothing great; i just C-M-x the define-module form, and then evaluate the test i'm interested in <civodul>i'm all ears if there's something better <davexunit>civodul: I find that I have to evaluate the test-begin and test-end first before I can get good output. <davexunit>I guess I don't know how to distinguish a pass from a fail without doing that and thus enabling the log output <civodul>if i know the body of the test should return #t, it's easy to just evaluate it and see if it does <davexunit>well thanks for letting me know how you do things <davexunit>not that srfi-64 is in guile I'm trying to make use of it and be better about unit testing my code. <davexunit>my old job was heavy on the TDD which drove me crazy, but unit testing is still important and I should still do it. :) <mthl>civodul: i think i misunderstood the meaning --localstatedir <mthl>i have put the value /usr/local/bin but i think i wasn't the right choice <ecelis>mthl: --localstatedir usually is /var <ecelis>--prefix=/usr/local maybe is what you want, which is the default value <mthl> thanks for the explanation <civodul>mthl: the Autoconf manual describes localstatedir as "The directory for installing modifiable single-machine data." <civodul>here that's mostly the store's database