<zacts>I still have one box running debian, I'll let you know the specific errors I get the next time I try it again. <zacts>from what I remember, it couldn't find guile for some reason <davexunit>so I have added ~/.guix-profile/bin to $PATH, do I need to modify $LD_LIBRARY_PATH as well to point to ~/.guix-profile/lib ? <davexunit>I have an SDL example program that I can't quite get to compile using the right libsdl <civodul>davexunit: normally no, you don't need to modify $LD_LIBRARY_PATH <civodul>just set $PATH and whatever "guix package --search-paths" says <davexunit>I should note that I'm using the pre-inst-env <davexunit>dave@izanagi ~/C/sdl> gcc -lSDL -I ~/.guix-profile/include/ -L ~/.guix-profile/lib test.c -o test <davexunit>/usr/bin/ld: cannot find crt1.o: No such file or directory <davexunit>/usr/bin/ld: cannot find crti.o: No such file or directory <davexunit>crt1.o is in /usr/lib/x86_64-linux-gnu/crt1.o but obviously I want to be using the guix paths. something is very wrong with my guix setup. <Steap>davexunit: does 'which ld' return the path of ld in ~/.guix-profile/ ? <Steap>hum, then /usr/bin/ld must be hard-coded somewhere <davexunit>yeah I'm not sure what is going on with my setup. <Steap>Does gcc call every subprogram in /usr/bin ? <viric>with -B you can override the hardcoded path for sbprograms <a_e26160>jmd: Concerning the license of xf86-video-dummy you wondered about yesterday: <a_e26160>Copyright 2002, SuSE Linux AG, Author: Egbert Eich ***a_e26160 is now known as ae237859
***ae237859 is now known as a_e
<jmd>a_e: Well, it doesn't say. <a_e>The default if it does not state a free license is to assume that it is proprietary. <jmd>Well yes, that is safest. <jmd>Did you email the copyright holders / file a bug report to ask what the license was? <a_e>No. I got the impression x.org does not care so much about freeness - there are tons of non-free fonts in the distribution. <a_e>And I do not think the dummy file is really important. If it is needed, we can worry about it. <jmd>I suspect the lack of license is simply that someone forgot it. <jmd>Somehow, in Debian they have put a free licence on it... <jmd>I wonder if that means they got permission, or simply copied it blindly? <a_e>I see, they used the X11 license. No idea where they got this from. Normally, they are very thorough. <jmd>I suspect they said "Comes from x.org; therefore X11 license." <jmd>The reason I asked about it, was that I was trying to package gtksourceview <jmd>two tests fail, apparently due to lack of an X server. <jmd>But if the copyright is held by SuSE Linux AG, then that's who we would need to ask, not x.org <a_e>Could you try to add xorg-server as an input? It contains xvfb. <jmd>Huh? I thought xvfb was built from xf86-video-dummy <a_e>No, it is inside /nix/store/...xorg-server-/bin/ <jmd>Then I totally misunderstood it. <jmd>Hey. So far as I can tell "SuSE Linux AG" does not exist. <jmd>There is a "SuSE Linux GmbH". <a_e>Maybe it was an AG at some point in time? See: <jmd>It seems that the pkg-config misbehaves when there is a very long PGK_CONFIG_PATH variable :( *jmd takes a rest from debuggin pkg-config <jmd>Well, it seems that pkg-config misbehaves when there is a a very long PKG_CONFIG_PATH <jmd>... which unfortunately is likely to happen in guix ... <jmd>It seems that although it works correctly with a line like pkg-config --modversion foo <jmd>it erroneously returns non-zero when invoked as pkg-config --exists foo <jmd>which is what happens in autoconf scripts. <jmd>Thus, ./configure was complaining that a package was not installed, when in fact, it was. <mark_weaver>jmd: can you give us a precise value of PKG_CONFIG_PATH for which 'pkg-config --exists <foo>' fails? (and tell us <foo> also) ? <jmd>I can pastebin the config.log if you want. <civodul>it's happening in a chroot build environment, right? <jmd>PKG_CONFIG_PATH is 2301 chars long. I suspect it may be overflowing a buffer or something. <jmd>Yes. when I do guix build <jmd>The erroneous result is on line 860 <civodul>i don't see anything wrong, except that libgoffice is clearly missing <mark_weaver>yeah, I have no idea why he thought line 860 had to do with pkg-config. <civodul>mark_weaver: any idea about the mit-krb5 test failure? <civodul>apparently one of the tests spins forever <mark_weaver>where is the hydra log? (I have no idea how to find things on hydra) <civodul>the important thing is "timed out after 3600 seconds of silence" <mark_weaver>I notice that there's a newer version of mit-krb5 than the one we have. maybe we should try upgrading before investigating further, in the hope that they might have fixed it.