pkgsrc-Bulk archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: pkgsrc-current-destdir Linux 2.6.24-1-686-bigmem/i686 bulk build results 20080716.1221

 >> > Then just remove the section, please.  Thanks.
 >> I don't see any reason for this, sorry.  This section shows how many
 >> packages including dependancies are broken per package maintainer.
 >> And this information is very useful.

> That section only shows that some people maintain more packages than
> others.

No. This section counts a number of broken packages per maintainer.
A number of packages maintained by individual maintainer
is another count. These two things are somehow interrelated but
in general they are completely different.

> Try this little exercise: count the number of packages per maintainer in
> pkgsrc, spread the failure % across this list based on the % of packages
> the particular maintainer owns and compare it with the list you produce
> right now.  I seriously doubt the list will differ significantly.
This is just another statistics about pkgsrc.
It may also be interesting.

> We don't believe in culture of blame in pkgsrc (and NetBSD).  People
> spend their free time for the benefit of others.  By composing such
> blame lists you can only discourage people from maintaining more
> packages because they would inevitably end up on the list.

First of all, I'll replace "Top offenders" with "Per maintainer
statistics" in the next version of distbb. The word "offender" is
probably not very appropriate. It was obviously migrated from Hubert's
bulk build and pbulk where it was used for packages.  This
word was probably a cause of misunderstanding. Note that english is
not my mother's tongue and if "offender" sounds offensive, I'm sorry
for that and I'll remove it from distbb logs.

Second, I'll try to fix these counts in future versions of
distbb. Currently packages marked as "not for Linux" are counted just
like all others.

Third, I'm OSS user for about ~9 years. And for more that 6 years I
develop small/medium-side open source software. For this period I
wrote lots of patches,fixes etc. for different programs and OSes. Ask
google about them. Though, open source and UNIX-like OSes is just my
hobby. I know exactly how hard development is and I don't blame my
colleagues. Yes, believe it or not, I think NetBSD and pkgsrc
developers are my colleagues.

Fourth, please, have a look at gnats and see how many PRs was sent by
me and how many bug and change-requests were sucessfully fixed and
implemented both for NetBSD and pkgsrc. Note that there were fixed
VERY serious bugs in NetBSD system and pkgsrc. When you do this, I
hope you'll understand that I'm rather active NetBSD/pkgsrc user and
I'm really interesting in doing NetBSD and pkgsrc better. This is
another reason why I DON'T BLAME YOU, NetBSD developers.

I use NetBSD at home and pkgsrc both at home and work because I share
NetBSD's and pkgsrc's goals: stability, good design, clean code and
portability. While sending all my PRs and seeing how they are fixed,
I've found that I like how fixes was made, almost always PRs I
reported were fixed correctly and very accurately. Many times(!!!)
I was surprised how meticulous and careful NetBSD developers are.
In most cases final solutions was not worth than that I suggested.
Sometimes people found bugs in my patches, corrected them and point
me out. Many times! As a result I said "I like it". No, I'm not
with NetBSD since its beginning, I use NetBSD and pkgsrc
just about two years or so.
I repeat, almost always NetBSD depelopment I've ever seen was VERY good.
Though, in most cases this was rather small parts of code, just fixes
and minor improvements.

Of course, there are things that I don't like in NetBSD and pkgsrc.  I
don't like words like "You just blame us, just do what I/we say
because I'm ... while you are not".  I don't like how you discuss
serious problems in mailing lists, that is I don't like arguments like
"Period", "We don't need this because we didn't need it for 10 years",
"haha, lemming's principle", "xxx is bloat, we don't need it". I also
don't like things like "Linux idiots", "Idiotic GNU" etc. said by
NetBSD developer.

Of course, I understand that I'm not diplomat too. I'm researcher and
software engineer. Sometimes I'm rather agressive. In general, I'm
straightforward person. If I don't like arguments like "nawk is
the only true awk" or "there were handwaving, and nothing else" I
just say "folks, I don't buy your arguments". If I treat your solution
as bad I say "I disagree, your solution is bad". If I dislike pbulk I
just I just say "pbulk works badly, was designed badly and is
maintained badly. This is the reason I wrote distbb".
This is my position. Some of you don't like Linux
and this is also your position. Right? And I assume you know what you are
talking about. I'm not system programmer at all. I program applications.

What is all this lyric about? I just want to say that neither of us is
ideal. And if you want I do something, please, show your arguments, first.
If you say "just do this, please", my first reaction is "No, I don't
see any reason for this".  What's wrong? Currently you just say "you
blaim us, your distbb logs are offensive", but actually YOU BLAIM ME.
Believe me, your reaction on "Per maintainer statistics" is totally
wrong. It is not blaim-list.  Just for comparison, see NetBSD official 

There in section "Severity" you'll find many NetBSD developers with a
number of critical/serious/non-critical bugs. Is this list offensive
too? Does this list blame NetBSD developers? Does GNATS blame NetBSD
and pkgsrc developers? I believe NOT. Then what's wrong with distbb?
I just did the same things! Absoletely the same as GNATS and many
other bug tracking systems do.

Now, lets look to this page

"Top Ten Offenders" section says that because of graphics/MesaLib and
lang/python24 failures maintained by bjs@ and recht@ more than 1500
packages failed. MAINTAINED by ...  Is this section offensive too? NO?
But why?  Just because I calculated summ of failures for each
developer? This is non-sense.

I repeat again, your reaction on distbb statistics is totally wrong.
If you were thoughtful, you can understand that this type of statistic
is VERY useful and may be very helpful if you think about it. You said
that some of developers maintain huge amount of packages. Yes, this is
true. And pkgsrc is not alone packaging system where a few developers
maintain more than hundreds of packages. One another example -
ALTLinux, developped and used mainly in Russia, though there many
developers work full-time. How does your remark correlate with my
counts (failures per maintainer). See below.


Colclusions? If I were you I'd like to say the following:
1) support for all these 14 platforms are not at the same high level
   as it is for NetBSD and for DragonFlyBSD.
2) Support for Solaris is terrible.
2.1) Support level for HP-UX and most others is unknown
2.2) Support for Linux should be improved
   If, for example, your packages failed under Linux I PERSONALLY
   can help you to test it
3) pkgsrc needs more bulk build results in pkgsrc-bulk@ mailing list
4) pkgsrc PACKAGERS need help from people running Solaris, Linux, HP-UX etc.
   They (me!) can help improve support for foreign systems
5) pkgsrc need more packageRS, see distbb statistic about
   pkgsrc-users@ pseudo maintainer
6) More active work with upstream is necessary.
   NetBSD, Solaris etc. are not mainstream platforms.
   Most developers use Linux and many of them just don't know anything
   about portability.
   They don't know anything about NetBSD in particular.
6) millions on monkeys cannot help if you need stability.
   Code readers are needed. Code review is needed.
   Persons that READ the code are much more important
   than those who WRITE it. Theoretically, OBSERVERS campaign may help.
   Appropriate software/per package mailing lists, for example, may help.
7) pkgsrc need better PR
8) The fact that some pkgsrc developers maintaine hunders of packages
   is actualy a PROBLEM of pkgsrc. This is not normal for packaging system
   running 14 software platforms.
9) Ideal situation for pkgsrc is when EVERY package has at least 14
   maintainers (not pkgsrc-users@) with single "leader".
8) etc.

You've made completely different colclusion. You just said "you use
our mailing list to blame us, you don't do things NetBSD developers
say you to do, perhaps you'd like to re-think your actions" etc.
I don't understand your position.  Believe me, such position can improve
neither NetBSD not pkgsrc. This is the way to loose your users and
(who knows) future developers.

"Perhaps you'd like to re-think your actions".

Best regards, Aleksey Cheusov.

Home | Main Index | Thread Index | Old Index