IRC channel logs


back to list of logs

<mark_weaver>civodul: ooh, I see hydra building libtool-2.4.4 :)
<mark_weaver>thanks for working on it!
<civodul>you're welcome :-)
<civodul>good night/day!
***svetlana is now known as gry
<Steap>Is there a procedure that flattens a list in Guile ? Can't find it.
<Steap>oh, there is "(fold append '() l)"
***gry is now known as svetlana
<mark_weaver>Steap: it's called 'concatenate'.
<mark_weaver>'append' on two arguments takes time proportional to the length of the first argument, so 'fold' will actually be sort-of quadratic. 'fold-right' would be better, but simply calling 'concatenate' is best.
<civodul>Hello Guix!
<civodul>iyzsong: do you want Hydra to build wip-qt?
<civodul>or did you want to make additional changes before?
<svetlana>Hello civodul (and everyone else)! Happy new year, and happy new name. :-D
*davexunit seems to have hosed his debian testing system. curses the lack of transactional rollbacks.
<davexunit>I can't use the sign procedure in (guix pk-crypto) module because it throws the error "invalid object" when I try. I don't know what's wrong with the libgcrypt on my system. :(
<civodul>davexunit: does "make check TESTS=tests/pk-crypto.scm" pass?
<civodul>there were problems with older libgcrypts, i think
<davexunit>yup, that passes.
<civodul>davexunit: could you post code the actual code?
<civodul>it may be that 'sign' receives an ill-formed sexp or something
<civodul>it's easy to get it wrong
<civodul>the first argument must be a canonical sexp as returned by 'bytevector->hash-data'
<davexunit>ah, okay. thanks.
<civodul>see the "sign + verify" test in tests/pk-crypto.scm
<civodul>yeah, easy to get wrong :-)
<davexunit>I've tried many permutations and couldn't get it right.
<civodul>the gcrypt manual has all the details, but we should probably improve the docstrings at least
<civodul>let me know if you find other things like that
<davexunit>okay. will do.
<davexunit>if I can get signing working, I should have a patch to post for 'guix publish' soon.
<davexunit>it's hopefully the last thing that 'guix substitute-binary' will complain about
*davexunit heads to the store
<nkar>civodul: unfortunately, I don't have much free time these days, and I don't expect this to change in the near future. I gave up on packaging coq and ocaml, for instance. so, if someone else wants to update the logo, that'd be great.
<civodul>nkar: ok!
<civodul>could you send your status for Coq and Caml?
<civodul>it's always a pity to lose hours of work
<nkar>well, I haven't done much
<nkar>I just bumped the versions and tried to build without tests to see if it works at all
<nkar>it seems that the ocamlfind program is need for coq to build
<nkar>but I haven't tried to package it
<civodul>iyzsong: your gmail address rejects messages from me :-/
<nkar>civodul: no spf or dkim?
<civodul>it's gmail itself that considers my message as possibly unsolicited
<civodul>davexunit: did you get signing to work BTW?
<davexunit>civodul: haven't had a chance to try again yet.
<davexunit>I've been mostly AFK
<davexunit>cleaning and cooking.
<civodul>which is important too! ;-)
<davexunit>let's give it a try now.
<davexunit>I have a few minutes before I have to check the oven again.
<davexunit>civodul: the docstring for bytevector->hash-data says that the bytevector should contain a hash, but that might not be the case for me, for some definition of "hash".
<davexunit>the way the Hydra perl code signs a narinfo, is by passing in all of the other narinfo text to the signString function.
<davexunit>(pardon the extraneous comma)
<nkar>civodul: my educated guess is that gmail uses several indicators to mark a message as unsolicited
<civodul>davexunit: yes, but technically, it first hashes it, and it's the hash that is signed
<civodul>i think(?)
<nkar>civodul: there should be a link in the bounce message explaining these triggers
<civodul>davexunit: yeah right, like in the "sign + verify" test
<nkar>civodul: such as the lack of the spf record or dkim signature
<nkar>but there may be other reasons. only google can say for sure
<civodul>there's a link to a web page that doesn't say much
<nkar>well, when I was getting these bounces, there was a link with the aformentioned reasons, among other things
<nkar>did you attach anything to your message? it might be a trigger, too
<nkar>i'd try searching the web for similar reports
<nkar>or ping a googler if you know one
<davexunit>civodul: thanks! got
<davexunit>got 'sign' to work
<civodul>i think we should do a 0.8.1 this month, no?
<grasshopprWhoppr>would you need some point-release testing?
<civodul>probably, yes
<grasshopprWhoppr>I'd like to try on an old laptop
<davexunit>civodul: so, I've signed something, but I'm not sure how to render it.
<davexunit>hydra base64 encodes the signature. do I base64 encode the entire sexp that is returned from sign?
*civodul checks
<civodul>yes, see 'signature-body' in tests/substitute-binary.scm
<civodul>and you actually sign the hash of all the narinfo up to and not including the 'Signature' field
<davexunit>I managed to figure out that much :)
<civodul>none of this is particularly well documented as you see
<civodul>nkar went through all this before :-)
<icarious>which image should I download to install guix in KVM ?
<icarious>the USB image comes as .xz .. can I install it using KVM?
<davexunit>icarious: yes, just extract and boot the image.
<icarious>davexunit: ok thanks. downloading it
<icarious>meanwhile m checking the package list :D
<davexunit>feel free to package what you find missing ;)
<icarious>does the system include non-gnu packages? or its a pure gnu system
<svetlana>probably bind9 if it's not there already
<svetlana>it does include non-gnu packages
<davexunit>icarious: plenty of non-gnu packages.
<davexunit>we accept all free software :)
<icarious>so this is going to be the de-facto official GNU system :D
<nkar>davexunit: if you still have questions, I suggest to search the mailing list. there should be some info on the signature process
<davexunit>nkar: thanks. I *think* I have it.
<davexunit>going to see if 'guix substitute-binary' is cool with it.