IRC channel logs

2014-07-14.log

back to list of logs

<DusXMT>civodul: okay, I can boot the system from install.scm if I switch the hard drive into IDE mode. Now I wonder, why did my own operating system declaration didn't worn in this case? (it kernel panicked - tried to kill init)
<civodul>dunno
<civodul>could you post your declaration?
<DusXMT>sure
<DusXMT>if I can get this thing to boot properly...
<DusXMT>civodul: http://pastebin.com/tZxvBHG3
<DusXMT>I pretty much just copied it off the info page, only editing some rudimentary things like the hostname, partition name, time zone, etc.
<jmd>So I have a build which appears to have completed successfully:
<jmd>starting phase `strip'
<jmd>phase `strip' succeeded after 0 seconds
<jmd>
<jmd>But then I get:
<jmd>builder for `/gnu/store/8jl54px08x60g32prccqjxig8gkx41ka-qcad-3.6.2.0.drv' failed to produce output path `/gnu/store/dajx64i35lb0gf6shj9n52s0c3yw7i0m-qcad-3.6.2.0'
<jmd>@ build-failed /gnu/store/8jl54px08x60g32prccqjxig8gkx41ka-qcad-3.6.2.0.drv - 1 builder for `/gnu/store/8jl54px08x60g32prccqjxig8gkx41ka-qcad-3.6.2.0.drv' failed to produce output path `/gnu/store/dajx64i35lb0gf6shj9n52s0c3yw7i0m-qcad-3.6.2.0'
<jmd>killing process 16803
<jmd>guix build: error: build failed: build of `/gnu/store/8jl54px08x60g32prccqjxig8gkx41ka-qcad-3.6.2.0.drv' failed
<jmd>
<jmd>Why did it fail to produce the output path?
<civodul>sneek: later tell jmd "failed to produce output path" means that the package's build system did not actually create and populate $prefix
<sneek>Okay.
<davexunit>failed to resolve partition label: gnu-disk-image
<davexunit>:(
<davexunit>I thought that my i686 boot image would finally work
<civodul>Hello Guix!
<mark_weaver>Hi civodul!
<bavier>hello!
<mark_weaver>civodul: if you want to try to connect the Loongson 3A to hydra, I can spare some time now to help debug.
<jmd`>Does guix have a patchelf utility like nix ?
<mark_weaver>jmd`: yes.
<mark_weaver>same program, although ours is patched to work correctly on MIPS.
<jmd`>How do I use it?
***jmd` is now known as jmd
<mark_weaver>jmd: search for 'patchelf' and 'augment-rpath' in gnu/packages/*.scm for examples. e.g. python.scm
<mark_weaver>you'll also need to make sure that (guix build rpath) is included in the imported-modules. See the #:modules and #:imported-modules arguments for python-2
<civodul>yeah, it's (guix build rpath)
<civodul>right
*civodul goes look at the MIPS machine
<jmd>python.scm seems just to declare it as an input. I can't see where it is used.
<mark_weaver>civodul: guix archive tells me: "generate-keys: unrecognized option"
<mark_weaver>ah, I just need to remove the 's'.
<civodul>ok
*civodul goes afk a little
<mark_weaver>civodul: I sent you the new RSA guix signing key.
<mark_weaver>civodul: you might have gotten two copies of that email, and it's possible that only the second one was signed by me. something went wrong on the first attempt to send it...
<jmd>Well that was dumb!
<jmd>I was idly brushing the dust off my notebook, and inadvertently presses the poweroff button.
<mark_weaver>jmd: the call to 'augment-rpath' is what uses it.
<mark_weaver>that's a scheme procedure that runs the program.
<jmd>mark_weaver: Thanks.
<jmd>patchelf doesn't have an option to replace exactly one item in the path.
<waxysubs`>jmd: fwiw, it should usually be possible to arrange for the rpaths to be included during the original link, avoiding the need to patch it later (which has various problems, including not working properly with cross-compilation in the general case)
<mark_weaver>oops, that was me. I meant to type it here...
<jmd>Yeah. Possible, but rather difficult.
<jmd>I wish there was a substitute* which took its strings literally
*davexunit is re-building the i686 install image
<jxself>And then it will work?
<davexunit>I hope so.
<davexunit>I've got an x60s to attempt the install on.
<jxself>Thank you FSF.
<davexunit>yes indeed.
<davexunit>damn it, kernel panic.
<davexunit>In procedure open-file: No such file or directory: "/gnu/store/...-system/boot"
<davexunit>seems the image is bad, perhaps I didn't do the right thing when flashing the image.