Summarizing the big idea
Submitted by Bram Schoenmakers on 16 December, 2007 - 19:54.
My previous post about closing all KDE bugs generated a lot of feedback, more than I expected. The poll tells us that 63% of the people would like that idea and 37% of the people thinks otherwise. So there seems to be more support for this (crazy, controversial) idea than I initially thought.
But to start with, let's wipe away some misinterpretations on the article:
- "Oh no, KDE3 will be unsupported!" Definitely not true. Bugfixes which still suit in the old codebase still happen and will happen in the future. But the main focus is of course the KDE 4 platform.
- I have never assumed that having KDE 4 makes all bugs vanish all of a sudden. That's why I suggested reopening those which are still valid.
- We won't lose all valuable information which resides in Bugzilla. Closing is not the same as deleting.
- I didn't intend to touch the wishes, I thought I stated that pretty clearly in my post.
- I'm not going to summarize the replies mentioning Microsoft practices here. Too lame.
Besides the rather black and white approach of closing all bugs, numerous alternative solutions were suggested:
- Check for every bug entry in Bugzilla if it is still reproducible in trunk and close it if doesn't. I hope that suggestion was a joke.
- Have two Bugzillas: keep the current one with and start a new one with all KDE 4 stuff. I don't think this is a good idea: it will be overly confusing for both users and maintainers. Both groups have to keep track of two Bugzillas.
- Will Stephenson suggested to have a two week squashing period somewhere in the first months of 2008. Can we expect numerous people to clean up the mess in their component?
- Only apply my initial suggestion on unmaintained components in Bugzilla? But then we have to #define 'unmaintained'.
- Ask for more volunteers. OK, that's not too hard. Will you help?
- This one appealed me the most: the ping-close option. Close the bugs with no response after x months. The only problem is: who's going to implement it? Do we push this task to the overloaded sysadmins running Bugzilla? For myself, I wouldn't even know where to start to get this in Bugzilla. Maybe an intermediate solution would be adding a new status to Bugzilla like PING. And then check after x months which bugs still have that status and then mass-close them. That's more accurate than blindly closing bugs.
I think the last solution is a nice compromise for my initial idea and the idea of doing nothing at all. Let's see if this can be realized somehow.
Actually the discussion on
Actually the discussion on KDE-lists has shown that the idea of closing ideas was rather stupid and it has therefore been more or less turned down. For having posted such a question in a biased expression, 63% is actually rather poor. Finally, you should have diskussed this first before making your very strange "poll". To sum up, we should forget about this stupid idea as soon as possible and according to what I have read, this is exactly what already has happend.
Well, I disagree with the
Well, I disagree with the anonymous person above. Unless someone can come up with another realistic idea to clean up bugzilla, sending a message to everyone who filed a bug to ask them to check if it is still valid - and close it if they don't reply - that's imho the smartest thing we can do. Sure, many areas in bugzilla are relatively well maintained. But many are not - and I think Bram is someone who knows that better than most of us. Ignoring him or calling him stupid is... well, rather stupid I'd say.
Bram, I fully support you on this. There might be need for an opt-out thing for those apps which have a proper bugzilla management, but for the rest - this sounds fairly reasonable.
rinse and repeat
While I agree with you that we need a solution, I think its also very important to not keep on pushing an already shot down idea as you'll just alienate people.
On the kde-core-devel list I asked for some more statistics to make sure we actually know what we are talking about. This has been too much work, apparently.
I can come up with a dozen ways to make the load get less on bugzilla, but coordination is the prime one. And tools to help there are a close followup.
In short, I'm a bit disappointed that Bram just keeps on pushing his idea while a lot of people are misinformed (its not actually going to be a silver bullet) or are just plain not listening to the experiences of people working on the bugs are saying.
Constructive suggestions
* make a list of all bugs that actually need to be looked at.
Then let people look at it and either reassign it, close it or mark it as 'looked at'.
This means all bugs will at least have been checked for duplicates etc at least once, and not different bug-triagers won't go over the same bugs. Plus you'll have a much better idea of the state of bugzilla after you are done.
* All bugs that are reported against an older version should get a tag saying if they are still reproducable in kde4. Yes, its hard work, its manual work. But do it in combination with the above suggestion and you can spread the work out amongst the thousands of people that like to help.
See, the idea to remove the work by throwing out potentially good information just sucks IMO, a much better strategy is to make sure we fix the real problems that we have now. Which are
* The inflow of bugs is larger then the speed with which we can handle this. If the amount of bug triagers doesn't go up quite a bit, this will only get worse. So closing bugs doesn't actually solve anything.
* There are applications that have a huge amount of bugs open but never actually close them. In other words, they don't have a maintainer (at least for the bugs). Identify such applications and either assign a maintainer or be brutally honest and don't accept new bugreports (which hopefully will quickly get a new maintainer)
* Application maintainers that do maintain their bugs (like I do for KWord) still allocate 600 bug-ids, which looks a lot to some but its entirely acceptable for others. The problem here is that there is no way to figure out if those bugs are a mess or are neatly triaged.
Re: rinse and repeat
> On the kde-core-devel list I asked for some more statistics to make sure we
> actually know what we are talking about. This has been too much work,
> apparently.
Yes it was.
> In short, I'm a bit disappointed that Bram just keeps on pushing
I wonder if you actually read it. Or maybe I wasn't clear enough with the last suggestion: a ping-and-close would be an excellent alternative. But that burdens the sysadmins.
> * make a list of all bugs that actually need to be looked at.
Which are those 15000+ which I mentioned before. See my next point.
> But do it in combination with the above suggestion and you can spread the work
> out amongst the thousands of people that like to help.
Thousands? Come on, don't be that optimistic. A regular bug day which we held last year attracted a handful of people (about 30) despite the press given to it.
Furthermore:
I don't agree with closing the gates for unmaintained apps, because someone has to maintain that list. And I seriously wonder if closing existing bugs is worse than blocking bugs at all. As a user I would feel more desperate than having a closed bug where I have the power to reopen it.
I think it wouldn't even
I think it wouldn't even need to be implemented in Bugzilla. It's just some database queries that can be done from everywhere. Bugzilla afaik logs property changes (changes to fields in the bug report) with datetime fields, so activity can be judged by that or the last comment or some combination of both.
You don't need a 'PING'
You don't need a 'PING' status.
Mark all KDE1/2/3 bugs untouched 'x' months prior to the release of KDE4 as "Pending User Response", send an email to submitter & CC list, and then automatically close all bugs after 'y' months as 'Closed - No User Response' if there is no response from the submitter or those on the CC list.
technically feasible. politically? unknown. Benefit? only future can determine if so implemented.
good luck.
You don't need PING
In my opinion you don't need a status PING, you can use UNCONFIRMED.
About using UNCONFIRMED
The problem is that people don't always confirm bugs. It's too unreliable to base your actions upon it.
Too many bugs < too many projects
Kde has far too many projects. Some projects duplicate others. Instead of fixing Konqueror (as a file browser), we have Krusader and now Dolphin. Three projects with bugs where one without bugs would be welcome.
Krusader and Dolphin are
Krusader and Dolphin are both very different from Konqueror. A lot of people, especially those that used norton commander back then, like to use a classical twin panel browser. Remember it's all about choice. Bug reports of the file part of konqueror now belong to Dolphin btw, because it's actually the same part. So no worries here.