Tim Retout's www presence

Mon, 16 Nov 2009

More RC bugs

Heh, thanks zack for your welcome. :) I'm afraid I don't make bug fixing look as easy or as organised as you do.

Some cheating to finish off the weekend: #548860 was reopened accidentally, #555898 was already fixed in another package, and #555939... actually, that did need a fix, but I just munged the test suite to expect some new error output.

Then I spent this morning writing half a testsuite for svn-buildpackage, and this evening just zoning out.

Sometimes I wonder whether fixing RC bugs is actually improving the quality of the release, or just letting us make a buggy release sooner. Non-RC bugs get pushed to the back of the queue. Hmm. But the pkg-perl team has around 170 packages with bugs in, and not that many people fixing bugs. I have to agree with corsac on the general point that just about everywhere in Debian could use more people.

There's a huge role for non-DDs to play in getting fixes into Debian, but as far as I remember, the emphasis of the mentors documentation is on packaging rather than bug fixing. I was barely aware that I was even allowed to prepare an NMU. And from what I know of the NM process now, a huge list of RC bug fixes would go a long way to establishing credibility as a potential DD - longer than a huge list of poorly-maintained packages.

It seems to me that the mentor relationship works better when DDs get to know particular people, and regularly sponsor uploads for them. This way they are in a better position to advise and sponsor any NMUs that might come along, and eventually to advocate them as an NM candidate. But it also seems to me that debian-mentors is not geared up for this at the moment.

To conclude, it's 1am, and I shouldn't be posting controversial thoughts on my blog. :)

Posted: 16 Nov 2009 01:09 | Tags: , | Comments (3)

Comments

Posted by btmorex at Mon Nov 16 03:54:57 2009
Other than the fact that Debian just has a really high barrier to entry for contributions, the biggest problem is that the packages that tend to be the easiest to make a small fix in are the same one's that have missing-in-action maintainers. So, even if I make a fix to a package (which I've done in the past locally), most likely I don't contribute that back to Debian because the maintainer of that package is no where to be found.

Sometimes I think Debian would be better off if packages just became orphaned if the maintainer didn't at least check in once every 3 months or so. At least then, non-Debian contributers would feel okay trying to find a sponsor to upload a fixed version of a package.

Posted by Simon McVittie at Mon Nov 16 12:03:12 2009
I was asked to fix and NMU at least three RC bugs (sponsored by my AM) during the New Maintainer T&S process, so at least some AMs are putting a bit of emphasis on that :-)

I've been looking at RC bugs recently (although not in a very focused way - maybe I should join in with zack's initiative, or maybe I should get on with the other million things I've promised to do instead) and noticed that there are a reasonable number where a non-DD has attached an NMUable patch (produced by nmudiff, no less).

However, I've specifically declined to sponsor at least a couple of those (with appropriate apologies to the patch author), on the basis that the fixes look good, but the package didn't seem like its presence improved Debian; the maintainer agreed, and asked for its removal. As our glorious leader points out, there's "too much crap in the archive" :-)

So, I think it's worth considering when investigating RC bugs in obscure packages: should we address this bug by fixing it, or by dropping the package? In cases where the package doesn't really add value and is likely to be a maintenance burden to the project in future, it can be more worthwhile to take that excuse to drop it, rather than keeping it limping along in a semi-useless (but not RC-buggy) state.

Posted by Tim Retout at Mon Nov 16 22:30:17 2009
@btmorex: For bugs marked 'important' or higher, there would be a good case for an NMU, and your update could get sponsored. For smaller bugs, I hope you attached your patch to the BTS. :) They should get fixed once there is an active maintainer.

But in my experience, one thing we are not short of is bugs to fix.

@Simon: Mmm, good point about the archive being full of crap. :) Dropping rubbish packages can't be a bad thing - this always comes up during BSPs.

Name:


E-mail:


URL:


Comment:


Anti-spam question: What is the colour of both (hydrated) copper sulphate and (anhydrous) cobalt chloride?

< November 2009 >
SuMoTuWeThFrSa
1 2 3 4 5 6 7
8 91011121314
15161718192021
22232425262728
2930     

Contact

Tim Retout tim@retout.co.uk
JabberID: tim@retout.co.uk

List of non-fail

Copyright © 2007, 2008, 2009 Tim Retout