Adam,
> May I suggest that you modify bugzilla so that the general public
can't
> submit P1 issues? A lot of this has to do with expectation, and there
> is a general expectation in the software industry that P1 issues
> usually get some attention within 24 hours (even in open source
> projects), simply because of their nature...blocking issues have
entire
> teams sitting on their hands, which you can appreciate is incredibly
> expensive.
as an ISV serving 1.200+ enterprise customers I can tell you that this
expectation is usual. Moreover, most of our customers expect that P1
bugs are getting fixed within few days even without having paid a
subscription, because the issue is so severe that others (paying
customers) are getting affected of it, too. As a result, our company is
very eager on fixing ANY bugs ASAP independent of the status of the
reporter. Even low priority bugs will get fixed not later than in the
next major release. So bugs have a higher priority for us than any new
feature. We had been very successful with this strategy over the past
20+ years and our customers appreciate it with the fact that some of
them are paying subscriptions for 20+ years now. I think this is what
people expect to find with an infra structure product like GlassFish,
too. It is just inacceptable that even for P1 bugs Sun is making a
difference whether the reporter is a paying customer or not. Only
severity should play a role.
On the other hand I see that the community is not as active as in other
open source projects and I would appreciate that people invest more in
the product. For me, I see my role in two areas: Distributing the
product all over the world through our customers, and reporting new bugs
collected at our customers. Just count all the bugs I reported so far
and take into account that in Q1/2009 we distributed GlassFish to about
50 enterprises so far. Yes this is not much. But it is very hard to
distribute a software that has a P1 since months. This is not worth
nothing, and Sun actually appreciated it several times. So a company
acting like our's should be taken as seriously as one paying customer,
shouldn't it?
Also I must say that it is very hard to contribute to the GlassFish
project due to two things: (a) The code is open, but the process is not
at all. People are used with the principles of democracy and it is not
nice to know that one shall contribute for free but one is not the least
involved in any architectural or management decisions. Sun just does
behind closed doors and the community has to accept from it. So this is
open source, but really just SOURCE, it is not an open PROJECT. At least
for me, this is a real problem, since often the problem behind a bug is
not the code itself but the architecture -- which is something
Sun/Oracle and only Sun/Oracle decides about, but not the community --
and all proposals one sends in are more or less unanswered; so the bug
cannot get fixed easily and will become a quirky hack (and before
anybody says, no the code is great: Please check this forum's archive
about the hangs I reported and the reasons explained by Sun: quirks and
bad hacks). (b) The code is very complex and needs long time to really
get understood. I am a contributor to TopLink / EclipseLink and must say
that even simplest things had been architectured so complex that it took
two Oracle people for several days to explain all the quirks you have to
remind for a simple feature done in one hour. If the architectural
quality of the code would be better it would be easier to contribute so
more people would do (talking about GlassFish as an umbrella, not the
core product itself, as I don't know that source). But this is something
that cannot be changed fast, and in the case of Oracle, there was just
no will to improve this "just" for the sake of more external
contributions. In fact, I more or less gave up after one year on
contributing, after Oracle screwed my contribution for the sake of their
existing test suite that itself was buggy but nobody wanted to fix...
(they understood that my code was correct but their test suite was not,
but it was more essential that all tests are made green with the least
effort - which was to NOT fix the test suite but remove my features).
So, the tip I could give Sun/Oracle would be: Make an Open Project, but
not just Open Source. Give up control. Elect the project leaders.
Discuss architecture in the forum. Do not decide anything outside of the
forum. THEN a lot more people could participate and THEN you have the
power of the community. Until then, GlassFish will be just a Sun/Oracle
controlled proprietary product that has it's source getting published
plus a discussion forum plus a (ignored-due-scarce-time) bug tracker.
GlassFish has a large community. Start to use that power.
To say it clearly: I really love GlassFish and the Sun/Oracle people are
really cool and helpful. This discussion is NOT about the product or the
people. I just want to point out where the problem is: In decisions
behind closed doors.
Regards
Markus