IRC logs for #baserock for Wednesday, 2016-08-24

*** rdale has quit IRC00:05
*** locallycompact has joined #baserock06:56
*** fay has joined #baserock07:10
*** fay is now known as faybrocklebank07:10
*** toscalix has joined #baserock07:49
*** rdale has joined #baserock07:53
*** anahuelamo has joined #baserock08:18
*** jonathanmaw has joined #baserock08:36
*** paulwaters_ has joined #baserock09:29
*** lachlanmackenzie has joined #baserock10:19
paulsherwoodany ideas what ' error: possibly undefined macro: AC_DEFINE' means?10:26
richard_mawprobably an old configure.ac10:26
richard_mawmay not be compatible with the latest autotools10:27
paulsherwoodthis is from master of bubblewrap10:27
richard_mawpossibly a too new for the version of autotools then10:28
* richard_maw has only seen it the other way around10:28
* paulsherwood wonders if there's a bubblewrap channel somewhere10:28
richard_mawyou could probably ask in the project atomic one10:28
paulsherwoodgood thinking10:28
*** anahuelamo has quit IRC11:00
*** anahuelamo has joined #baserock11:00
paulsherwoodwell, in any case, i'm wondering if means that bubblewrap wouldn't work for most users?11:06
richard_mawI think all those options were added by something like 3.10, so it's been a couple of years since they were made available11:09
paulsherwoodoh, ok11:10
paulsherwoodi'll keep exploring then11:10
richard_mawand even if you're not on 3.10, I know some vendor kernels have backported all those features11:11
richard_mawRHEL kernels are scary11:11
persiaOne of the best examples of preserving a version number while massively patching the code.11:16
paulsherwoodleeming: i'm seeing which reminds me of some things you fixed in ybd. is it possible sandboxlib has the same issue?11:37
richard_mawpaulsherwood: that's because python isn't designed to chroot11:38
richard_mawpaulsherwood: it needs to do some string encoding, but the code for that is a library that is loaded on-demand11:38
richard_mawso after it chroots it can no longer find it11:38
richard_mawI think Sam's work-around involved explicitly loading encodings before chrooting11:39
* paulsherwood guesses that upgrading sandboxlib will fix the problem11:41
paulsherwoodactually, would putting that same thing into ybd itself address the problem, or does it need to be in sandboxlib?11:43
* paulsherwood is hoping python wizards here may know11:44
richard_mawdoing that in either would work, it's more appropriate for it to be in sandboxlib than ybd11:45
paulsherwoodack. but i think that fix hasn't made it into the pip version of sandboxlib11:45
persiaIs there anyone who can do a release?11:47
persiaThat should be easier than trying to port the fix to sandboxlib clients11:47
paulsherwood is here, but I don't know who has the requisite pypi credentials11:47
persiaIf the issue is single-human-with-access-keys, a conversation with the PyPI admins to add someone else (with a history of commits) would probably go smoothly.11:47
paulsherwoodoh, ok11:48
persiaWhoever does it needs a username on PyPI, but those are useful anyway11:49
persia isn't a bad way to go, once something is relatively stable, and if there are sensibe acceptance tests11:50
persia(for projects already on github, etc.)11:50
* paulsherwood has a pypi username, but no commits12:28
leemingpaulsherwood, reading from your ping to here, sounds like pypi is out of date and we need to update the release12:30
leemingI could try to contact sam on a back channel? if he can add someone else as a maintainer/admin :\12:31
paulsherwoodleeming: yes please12:31
paulsherwoodI/O error : Attempt to load network entity
paulsherwoodit is only us who think that make files shouldn't head out to the internet?12:40
paulsherwood(this is me trying to build bubblewrap with ybd)12:41
persiaNo, but the other camp is also not concentrated.12:41
persia(at least for "us" interpreted as "baserock folk": if "us" is "folk that think builds should not rely on network access", then it is only us :) )12:41
richard_mawpaulsherwood: that'll be docbook trying to be helpful. it usually loads from a pre-populated catalog, using the urls as cache keys. This catalog is usually provided as a package so you don't need to fetch from the internet.12:42
persiaI think the oldest and largest concentration of folk that think builds should not rely on network access is likely in the Debian community.  They have been pushing that agenda for most of the life of Debian.12:42
*** locallycompact has quit IRC12:42
persiaAlso, they have lots of patches that allow things to build without network.12:42
paulsherwoodrichard_maw: any ideas how to get aroudn this?12:46
richard_mawyou need to populate the catalogue somehow. Since we need it to be a git repository, it may be worth looking to see if debian, or some other project has a catalogue cache in a repository12:47
paulsherwoodeek. no way to say 'don't bother with the docbook stuff' ?12:48
persiapaulsherwood: You can patch it away, but that usually means no docs, which is rarely what you want.12:49
persiaI'd recommend having a docbook-cache chunk, and having stuff build-depend on it, and then not including it in deployed systems.12:50
persiaThat's a close match to the common solution, and once you've set up the cache chunk, it should just work for new things that are encountered (rather than needing to patch each one)12:50
* paulsherwood wonders if just having the docutils stratum will solve this12:52
*** locallycompact has joined #baserock12:52
leemingpaulsherwood, message fired off to sam, but unknown timescale for response13:07
*** CTtpollard has quit IRC14:03
*** CTtpollard has joined #baserock14:03
paulsherwoodleeming: thanks. in the meantime i fixed it by adding tristan's hack into ybd
*** paulwaters_ has joined #baserock15:03
* paulsherwood is still wrestling with the path/name dichotomy in definitions15:37
locallycompactthe answer is things shouldn't ever name themselves15:46
paulsherwoodi don't see how that helps, sadly. people use names to refer to things. definitions has always contained names (although they've not been consistently used/enforced properly) i could cause ybd to ignore all of the names in definitions, at the price of making all of the logs harder to read, and causing all previous builds/artifacts to be obsolete16:06
*** jonathanmaw has quit IRC16:08
richard_mawpaulsherwood: "people use names to refer to things." I don't think locallycompact is arguing we don't.16:10
richard_mawyou could always trim the length of the filenames when logging them, though common suffixes might be a problem16:10
locallycompactthe parent container already gives its children names, the name: field in chunk and stratum morphs are entirely superfluous, the parent already declares a name for each of its members16:12
paulsherwoodi'll think about that16:21
*** faybrocklebank has quit IRC16:22
jjardoncleanup started:
jjardonpaulsherwood: ^16:38
* paulsherwood is looking at something similar for a downstream project. interleaving of upstream and downstream is still tricky :/16:42
baserockgerritJavier Jardón proposed baserock/baserock/definitions: strata/multimedia-gstreamer-0.10.morph: gstreamer-plugins-base@0.10 -> gstreamer-plugins-base-0.10
jjardonthat's the only problem in definitions master16:49
paulsherwoodjjardon: i think we agreed on the @ approach, hence gstreamer@0.10 - would you be ok to rename the file, instead?16:50
*** radiofree has quit IRC17:17
*** radiofree has joined #baserock17:24
*** radiofree has quit IRC17:28
baserockgerritMerged baserock/baserock/definitions: strata/network-security.morph: Update gnutls to gnutls_3_4_14
baserockgerritMerged baserock/baserock/definitions: strata/network-security.morph: Update nettle to nettle_3.2_release_20160128
baserockgerritMerged baserock/baserock/definitions: strata/network-security.morph: Update libtasn1 to libtasn1_4_9
baserockgerritMerged baserock/baserock/definitions: strata/network-security.morph: Upgrade nspr to NSPR_4_12_RTM
*** radiofree has joined #baserock17:44
*** toscalix has quit IRC19:35
*** locallycompact has quit IRC19:48
*** locallycompact has joined #baserock19:49
baserockgerritJavier Jardón proposed baserock/baserock/definitions: gstreamer-plugins-base-0.10.morph: Change name to gstreamer-plugins-base@0.10
baserockgerritJavier Jardón proposed baserock/baserock/definitions: Move gssdp.morph to strata/dlna-services/
*** baserockgerrit has quit IRC22:26
*** lachlanmackenzie has quit IRC22:29
*** lachlanmackenzie has joined #baserock22:31

Generated by 2.15.3 by Marius Gedminas - find it at!