Control: tags -1 moreinfo Hi Benjamin, On 06-05-2019 12:47, Benjamin Drung wrote: > Am Donnerstag, den 02.05.2019, 21:05 +0200 schrieb Paul Gevers: >> Control: tags -1 moreinfo >> >> Hi Benjamin, >> >> On Thu, 18 Apr 2019 13:01:31 +0200 Benjamin Drung >> <benjamin.drung@cloud.ionos.com> wrote: >>> This version fixes the test_xen_virtual test case (bug #922352) and >>> exposes tornado4 as tornado for zmq.eventloop.ioloop (bug #924763). >>> Our >>> salt 2018.3.3+dfsg1-1 package introduced a big patch to use >>> python3-tornado4 (instead of python3-tornado) due to missing >>> support for >>> tornado version 5. Without the fix for #924763, >>> zmq.eventloop.ioloop >>> will import tornado version 5 (if python3-tornado is installed). >> >> Both bugs have severity normal. Do you really want to bother now or >> is the severity not correct (then please fix that and elaborate)? > > Bug #922352 can cause a build failure (and does on Ubuntu). Therefore I > raised it to serious. > > Determining the severity of bug #924763 is more complicated. The > reporter stumbled over a warning spit out by salt. The warning message > by itself is more or less harmless, but the underlying problem of the > wrong import might have bad effects. I haven't seen any yet, but they > might be there. IMO we shouldn't release salt with an issue introduced > by one of our patches. Ack. >>> I also included fix-various-spelling-mistakes.patch which fixes >>> several >>> spelling mistakes. Because this patch file is long, I excluded it >>> from the >>> attached debdiff. >> >> Bugs can be introduced that way. I am not going to review that diff, >> fixing spelling mistakes at this moment isn't appropriate unless >> these mistakes are crucial somewhere. > > I can drop that patch again when this is the only blocker for getting > the unblock request accepted. Please do. [...] >> You didn't even elaborate on all the (at this phase of the release >> inappropriate) changes to the packaging. There is even a newer >> version >> than the one you already mention in a follow up in this bug. > > Which changes to the packaging do you refer to? The once you discuss below. > Running the tests with LC_ALL=C.UTF-8 fixes a build failure in case the > building machine uses an ANSII locale, which would be worth another RC > bug report. Ack. > Upload 2018.3.4+dfsg1-3 repairs the documentation in salt-doc. It > ensures that the pre-built minified Javascript and CSS files are not > leaked into the salt-doc binary package and that all created symlink > are correct. Before this version, salt-doc contained broken symlinks > and the search did not work. I am very uncomfortable with these changes at this moment. > systemd 241 broke salt. Upload 2018.3.4+dfsg1-4 fixes that. There were > no bug report opened for it, but it would be worth one RC bug. Ack. > The newly opened RC bug #928337 was fixed in 2018.3.4+dfsg1-5. Ack. So, all in all, I don't want to unblock the new upstream with your packaging, we're too late in the cycle and the version really doesn't match the freeze policy. However, I am offering you an upload based on the version currently in buster via t-p-u. If you go that route, please only fix bugs 919849, 928337, 922352, 924763, the LC_ALL=C.UTF-8 item and the systemd issue. Please fix 919849 without switching your build to sphinxdoc, that isn't appropriate at this moment. You can remove the moreinfo tag when you either have a debdiff ready, or if your confident, uploaded to buster. Paul
Attachment:
signature.asc
Description: OpenPGP digital signature