On 10/05/2015 04:57 PM, Scott Kitterman wrote: > I agree that disabling package test suites doesn't improve their quality.
That's not what I did, I blacklisted these unit tests which were failing, and kept all the others. As these unit tests were anyway broken, it doesn't mater much. > Were these bad tests? Did you report these issues upstream? Upstream will *not* fix the issue, because you know, they "fixed" it in their CI by adding an upper version bound in the pip requirements, which is fine for them in the gate. It is fixed in OpenStack Liberty though, which I will soon upload to Sid. >> It is also to be noted that mock is maintained by upstream OpenStack >> people (ie: Robert Collins), and therefore, should be released in Debian >> at the same time as other testing tools and the rest of OpenStack: >> testools, testscenarios, subunit, testrepository, and many more. So in >> the future, I'd advise to follow upstream release schedule. I would >> encourage, if you don't mind, to put mock into the PKG OpenStack team, >> because that's where it belongs. If we don't do that, and without being >> careful, then breakage is to be expected. > > I think this exhibits exactly the myopia others have complained about. > python-mock has over 200 reverse build-depends in the archive (python3-mock > has almost a hundred more). It may be used by openstack and maintained by > someone who also works on openstack, but it is, by no means an openstack > thing. I'm surprised here by reading these numbers. How exactly do you show a package's reverse build dependency? "apt-rdepends -b -r" doesn't work... I remember I did it for mock though ... :( > python-mock was first uploaded to the Debian archive in 2009. I believe > openstack was started in 2010. Unless your theory of python-mock involves > time travel, I don't think it's possible to make python-mock appear because > of > openstack. That's not the case. It just happens that Robert Collins >> In other distributions (Red Hat and Ubuntu), everyone is aware of this >> kind of issue before uploading, and this kinds of things don't happen. >> There's a set of packages, goals and results for which these >> distribution care about, so package updates aren't just uploaded blindly >> without first making sure there's no grave consequence. I wish we did >> the same. That's a way more important than respecting package ownership >> for uploading to Experimental when there's no other reverse dependency >> (and within a team ... glups!). I do see that this view isn't shared >> among the persons who were self-appointed as "team leaders" though. In >> the long term, this lowers a lot the overall quality of Debian, so my >> hope is that everyone realizes what's important. > > I'm glad to hear other distributions are perfect. That's of course not what I wanted to say. > You knowingly ignored the team norms and clearly have no regrets and would do > it > again. WHAT ? I mean ... WHAAAT ?!? Do I express myself so badly, that it leads to this? Never, ever, I wrote something like this. So let me state it once and for all. 1/ I have already expressed regrets for this upload. 2/ This upload is a *mistake* because I checked too fast packages.d.o and saw "oh, DPMT, let's upload..." when I should have been more careful and really check for the actual fields (which were matching the "do not upload before ping" rule which I knew about, and not the "this is team maintained, go ahead..." as I thought it was by looking too fast). This is what made me say that writing this in a policy wont change anything: mistakes can still happen, no mater how big you write the rule. If we wrote "DPMT-do-not-upload" as maintainer, it'd be less prone to mistakes, as it'd appear as so in the pacakges.d.o page and everywhere else. The "trick" with uploads / maintainers field is just too confusing and error prone. 3/ No, I wouldn't do it again... Is it clear enough now? Re-read my past post, hopefully, you will realize that this what I wrote in the first place. > Personally, even if the team was the maintainer of the package, I would never > just upload something without giving a ping to anyone who was active as an > uploader. I think it's just polite, even if it goes beyond what the team > strictly requires Which I did many times too. >> Yes, probably what I did wasn't the correct social way to do it, since >> Sandro doesn't like it. But it was technically right to do so. I was >> also shocked to read that it was bad for me to "care more about >> OpenStack". Yes, of course I do. As this is what my employers pay me >> for. Also because I spend countless hours on it, every day. But does it >> mean I don't care about anything else in the archive, and don't mind >> breakage of other components? Certainly not. And I expect everyone to >> have respect for the Debian archive as a whole, do correct transitions >> and so on (yes, transitions... also for Python modules...). > > My impression based on your unwarranted hurry to upload python-networkx and > your rant about python-mock is openstack is all you care about. Then you have a wrong impression. I was trying to express the exact same opposite thing: that I care for everything else, and that I expect everyone to do the same. > I would really like it if instead of continuing to point fingers at other > people you would engage constructively. Scott, it is my belief that you have a very wrong opinion of who I am. What you wrote here, quoting me, shows the exact opposite of what I was trying to explain. I really fail to understand how we've came to that. Maybe you could try to re-read me again, trying to step in my shoes and try to see the world from my point of view, rather than just seeing black when I'm showing white? Cheers, Thomas Goirand (zigo)