=== juliank_ is now known as juliank
=== JanC_ is now known as JanC
[03:33] <tsimonq2> slangasek: Should be good to go then, he uploaded as is with Conflicts instead of Breaks, going with what you said, now it just needs pushing through NEW by an archive admin... :)
[06:00] <slangasek> fossfreedom: hi, so I'm reviewing NEW packages, and I'm confused about the changelog rationale for budgie-lightdm-theme-base.  What do you mean by "self budgie-desktop builds using the Ubuntu Minimal ISO"?
[08:53] <fossfreedom> slangasek: quite a few people in the community have asked for the ability to create their own builds of budgie-desktop.  So rather that starting from the full ISO and deinstalling they are building from - I presume the netiso or similar - and wanted a simpler method to get a working desktop without pulling in the various customisations/artwork/themes/icon-sets etc we have in Ubuntu Budgie.
[10:12] <ginggs> tsimonq2: do you think we can revert renaming library packages for g++5 ABI transition in healpix-cxx?  Then healpy will become sync-able.
=== nelhage_ is now known as nelhage
=== psivaa_ is now known as psivaa
=== plars_ is now known as plars
=== FourDollars_ is now known as FourDollars
=== andyrock_ is now known as andyrock
=== Tribaal_ is now known as Tribaal
=== BenC- is now known as BenC
[11:39] <mitya57> slangasek, hm, the policy says a different thing
[11:39] <mitya57> Conflicts should be used … when two packages provide the same file and *will continue to do so*
[11:40] <mitya57> (emphasis mine). So not when one of them is being obsoleted completely.
[11:40] <mitya57> That case seems to fall under "when moving a file from one package to another" which is Breaks according to the policy
[12:04] <fossfreedom> on the off-chance ... anybody have any ideas why our daily builds are failing at the moment?  https://launchpad.net/~ubuntu-cdimage/+livefs/ubuntu/zesty/ubuntu-budgie
[12:09] <cjwatson> mitya57: Conflicts+Replaces is described separately
[12:09] <cjwatson> mitya57: https://www.debian.org/doc/debian-policy/ch-relationships.html#s7.6.2
[12:21] <mitya57> cjwatson, thanks a lot, now I finally understand it :)
=== JackFrost is now known as Unit193
[19:25] <teward> this may sound stupid, but how do I disable PIE / PIC in package rules for a given package?
[19:25] <teward> (nginx merge won't build with it)
[19:41] <hjd> teward: https://wiki.debian.org/Hardening/PIEByDefaultTransition possibly -no-pie (don
[19:41] <hjd> 't really know much about this)
[19:42] <hjd> I'm not sure how large the Ubuntu delta for this package is, but when merging from Debian, presumably that part built successfully with PIE?
[19:56] <teward> we haven't had a merge since 16.04
[19:57] <teward> just direct updates
[19:57] <teward> so... yeah...
[19:57] <teward> blurgh.
[20:00] <hjd> oh :/
[20:04] <hjd> I see, might be easier to look at whether PIE can be enabled again once the delta is more under control? :p
[21:22] <teward> it's not the delta, it's upstream changes that introduce breakage
[21:23] <teward> in dynamic modules
[21:23] <teward> so there's no solution at the moment
[21:47] <hjd> teward: I'm not familiar with the nginx package so I might be missing something obvious, but if the problem is in the upstream code, shouldn't that cause problems in Debian as well? Looks like it built fine there just a couple of days ago.
[21:59] <tzero> uscan is supposed to work, right?
[22:00] <tzero> like, `uscan -d` is supposed to follow the watch file and download something instead of report: "uscan warn: No upstream tarball downloaded."
[22:14] <hjd> tzero: Does it give any other warnings or error messages?
[22:14] <hjd> You could try with `--verbose` or even `--force-download` (check the manpage for details)
[22:17] <tzero> hjd: only "uscan info:    => Package is up to date for from" and "uscan info: Not downloading upstream package: "; it finds everything correctly (watch file is just the standard github boilerplate), mangles the download URL, then tells me it's not going to actually download anything
[22:17] <tzero> oh, force-download works.. why is that even necessary
[22:21] <hjd> I'm afraid I don't know.
=== foxtrot is now known as texas
=== texas is now known as foxtrot
[22:53] <ginggs> teward: try DEB_BUILD_MAINT_OPTIONS=hardening=+all,-pie in debian/rules
[23:09] <ginggs> autopkgtests against lmfit-py armhf fail since python-pandas is no longer available for this arch. Can anything be done to prevent these tests from being run, or the results ignored? affected packages are python-numpy,  python-defaults,  matplotlib, python-stdlib-extensions and pandas.
[23:12] <tumbleweed> ginggs: we can tell britney that lmfit-py is known to fail on armhf
[23:12] <tumbleweed> (force-badtest)
[23:14] <ginggs> tumbleweed: thanks.  anyone around who can do this? ^
[23:16]  * tumbleweed can
[23:19] <ginggs> would tumbleweed then, please?
[23:26] <tumbleweed> ginggs: it's not just armhf: https://launchpad.net/ubuntu/+source/pandas/0.19.1-3
[23:28] <ginggs> tumbleweed: see e.g. http://people.canonical.com/~ubuntu-archive/proposed-migration/update_excuses.html#python-stdlib-extensions
[23:28] <ginggs> only armhf shows a regression
[23:29] <ginggs> ah, we don't run autopkgtest on arm64 and powerpc
[23:34] <tumbleweed> ginggs: and the version that is in release is published for armhf, so the autopkgtests should be using that...
[23:39] <ginggs> tumbleweed: i think the python-pandas binaries in release have already been removed so it can migrate
[23:46] <tumbleweed> ginggs: aah
[23:48] <tumbleweed> ginggs: done
[23:59] <ginggs> tumbleweed: thanks! and goodnight