Re: Community projects having problems with infrastucture
 

I have been working on this a few times, together with merlin1991. Talked with Pali about it too a few times.

Last time I cleaned out the repo and database as far as I could, but it seems there is still a something lingering. This all started because the package maintainer changed numbering scheme a few times, triggering a really hard to find issue in version number comparison. I easily spent more than 2 days fixing all kind of related issues, but there is still a problem.

I've told Pali a year ago that he could fix the issue himself by just changing the name of the package. This way he didn't need to wait for any fixes, but he decided against that.

There is still a problem in the packages interface with this package and it is almost impossible for me to find and fix the bug. There is only one package in the whole repo affected by this, I feel we should not waste any more time on this and just cut our losses. Rename the package and be done with it. The time wasted on this can be better spent somewhere else.

Re: Community projects having problems with infrastucture
 

I've spoken with Pali a few times on IRC, he missed me a few times as he pinged me in the evening/night. I tend to be mostly online during central European business hours.

I can't recollect seeing any PMs or emails for that matter from freemangordon.

Haven't been reading the forums that much, just because of the sheer volume of messages/signal to noise ratio. Bootstrapping apps.formeego.org and harmattan related stuff took a lot of my time.

Re: Community projects having problems with infrastucture
 

I was trying to reach you on IRC, but iirc last time I have success was by the time we were discussing co-maintainership for CSSU repos(somewhen July-August 2011). Anyway, I really don't want to go into dispute who is right/wrong, I am glad to see myself out of the ignore list. Lets focus on the real problems.

One of them being kernel-power package name. There is rationale why Pali refuse to change it, there are just too many other packages depening on kernel-power, I don't think there is a way all of them to be actively maintaned so their dependencies to be changed accordingly. So a change in kernel-power will make all those packages incompatible with "kernel-power-new" or wathever fancy name we put on it. I am sure you can imagine the mess.

Could you please share the link to the code of the function which is responsible for package version comparison, after all we are developers dealing with much harder to find bugs than a glitch in some string parsing python/php/whatever function.

EDIT:

An ugly and nasty, but working hack, will be to add a code that checks for that particular package name and deal with it accordingly.

Re: Community projects having problems with infrastucture
 

Now that's what i could call "constructive" (both from x-Fade and freemangordon), in opposite to what "some other official guys" wrote earlier when community rep / infrastructure issues were mentioned (basically, just general blah blah).

Re: Community projects having problems with infrastucture
 

I tried this now in the importer. It imported the package now, but I can't guarantee that it won't get stuck somewhere else. Let's see if that is the case :)

I now special cased the 50 version, so it works for the current version. The hack would need to be updated next time though.

Re: Community projects having problems with infrastucture
 

So far everything looks OK(but lets Pali confirm), thanks. Anyway, I will ask for the code of the offending function again, so we can check if there is some kind of a workaround for the problem. I mean - is it possible to change the version for the next kernel-power in such a way so it won't trigger the bug.

Re: Community projects having problems with infrastucture
 

Sorry if that offended but that's the general tone I got from some statements and that was not aimed at you or anyone else directly. It really doesn't make any difference if one has been a member of the community a month or 6 years and that is no reason to judge anyone. At least I don't see "hey n00b" -culture here which is great. My motive was to try to get discussion going and getting problems resolved which seems to be working. I don't know why that didn't play out earlier but the important bit is that something is happening now.

Anyway. back to the QA thing.. So what would be the preferred way of cleaning the queue and making sure it doesn't get stuck so easily again? Automated promotions doesn't seem to be the answer but what about the other way around I proposed a bit earlier that if the the maintainer doesn't promote the package into extras within a month after the promotion is unlocked the package is automatically removed from testing?

Re: Community projects having problems with infrastucture
 

The main problem exists because we need to compare versions in MySQL. As is not possible, we create an index where we give the version numbers a certain weight. Here

Anyway, the problem might go away when the version in Extras doesn't have the old numbering scheme anymore.

If we ever move to OBS, we can leave this whole code behind :)

Re: Community projects having problems with infrastucture
 

Why would you remove the app from testing? If the app has passed the quarantine, has enough proper votes (both from Testers and Users) why won't you promote it?

If you do that, you won't have waiting applications in Testing but in Devel.

Re: Community projects having problems with infrastucture
 

True. What I'd like to see is the queue and it's pass through time getting shorter and I'm just throwing in ideas.

There where some good arguments against automated promotions so I proposed this. Personally I'm still more in favour of the automated thing since if a maintainer puts a package to testing and it passes QA it should be promoted. In case of "developer's remorse" of putting it to testing - it could be handled something like if the maintainer votes down the package then it's removed from testing regardless of what the other votes say. The QA quarantine time gives the maintainer some time to do this.

In either case the queue should be much easier to handle. I'm not saying that these are the only options. I'm sure there are other possible solutions out there.

Edit: It also might be worth considering that packages require at least N votes from supertesters or something to be promoted. This is to guard against getting the more popular packages through without proper testing