package shepherding

[Leonard den Ottolander]

Hi Alexander,

> Do you think we should be doing nothing but
> bug work? Any software *always* have bugs, so that would mean we'd
> *never* do any feature work or any other kind of important work.

This of course is not what I meant. I said spending time on bugs is
important. Not that you/we should spend all (y)our time on hunting bugs.
Squashing bugs is part of QA.

> A bugfree version of a 10
> year old piece of software. Nobody wants that. 

Then my name is nobody. But again, I am not arguing that we should not
improve on features.

> I don't remember the last time I
> actually had time to do a bugzilla query.

Here. That's what I mean. Shouldn't developers at least have some time
to spend on hunting bugs? A few hours a week? There have been occasions
where I reported serious issues (being crashes in important parts of
prominent software), and still the developer did not approach me for
more details but after some serious nagging on my part. Do I really have
to chase after developers on the mailing lists or in IRC if I already
filed in bugzilla? Hey I don't mind, but you should know that this is a
two way street.

> One could say its the fault of management that there aren't 10x as many
> engineers as we have now so we'd have time to fix more things.

Twice as many would be enough. But seriously, if you have just a few
people who can spend their time on hunting bugs that would already make
a *big* difference.

> However, I very rarely see anyone fixing any of the non-trivial bugs I
> own (and I own thousands of them, so I wouldn't mind some help).

Part of this problem is communication. You can't expect (most) outsiders
to fix difficult bugs in software that you are supposed to be the
authority on. This means that if you want others to make non trivial
fixes you will have to communicate some of your knowledge of the
internals of "your" software, or wait for someone to come around that
already has that knowledge.

Another approach would be to set up work groups of a few people
attending to a certain package. But that again is a management issue.
And it would still require the developer to have/take time to
communicate with these people.

> I may be a pessimist, but I doubt anyone would read any such docs.

Some people read docs. Then a lot don't. But you can't expect people to
read information that has not been written down...

All in all I am not arguing that all progress on developing new software
features should be stopped. But I do think bug hunting and squashing is
an essential part of QA. Developers should have/take (more) time to
either do it themselves or help out others who are willing.

Leonard.

-- 
mount -t life -o ro /dev/dna /genetic/research