Public SourceMage Discussion List

[Karsten Behrmann <BearPerson AT gmx.net>]

Okay everyone!

First, I would like to thank Unet and Seth for their kind words, the 
nomination
and the implied trust they put in me. I'm very honored.
I suppose there are several people who would vote for me, because they know 
me,
because I helped them in the past, because of my coding history, and for many
other reasons.

However, I think that I am a coder, not a coordinator. I honestly think
I wouldn't be that good at diplomatically rating approaches and communicating
results to people nicely. I can (try to) decide where my work is best put
(which is why I'm working on the installer and not my (uncertain-to-be-used)
ISO generation), but I'm not sure if I can make that decision for other people
and decide where to let them go their own way. I also don't like the extra
time I'd have to spend away from other stuff at TL duties ;-). I suppose [0]
sums up part of my point nicely.

Therefore, I think I should not and do not want to accept this nomination.
Instead, I would like you to consider David as a candidate.
He is my TL of choice, having nice communication skills and quite
some planning ability, as he showed in designing the new installer.

To give those of you who don't know it some background, I'll briefly recount
some history concerning the candidates in question:

I will leave out the episode with Hamish, who seems to have been an able
coder, because I was in the US on student exchange at the time and only
got it recounted second-hand.

I joined the ISO team in fall/autumn 2004. After talking a little with
Unet, I decided to write some sane ISO generation scripts from the
bottom up because I didn't like the existing methods I saw.

- I can't remember if I fully grasped benoit's scripts at the time, I think
I saw them as a disorganized heap - (IMO - I was used to afk's neat sorcery
code) unclean code with a lot of files that did not seem to have been used
for quite a while but were not cleaned up. -

Back on topic, I was trying to get my code to work right. I couldn't quite
see through the installer, it being a 1200-line-monster, so I didn't really
have a lot of knowledge of what files it needed and how they should look.
I created my first ISO, which some of you people tested, many thanks for
that. Bugs came out like maggots because of the installer, (missing)
interaction between the ISO generation and the installer, and lacking
experience on my part in ISO and especially cache file generation.

I remember one situation where I curiously noted that benoit "left" the
cauldron discussion channel instead of quitting the IRC server. I whois'ed
him, and came across a #benoit-iso channel. After it seemed to have been
decided that I should work on new scripts, benoit was quite
happily developing parallel to the community. His test ISOs weren't
being hosted on sourcemage.org like mine, but on one of his sites.
(for the record, it was [1] )

To add insult to injury, the installer he used on his ISOs was a fixed
version with quite some bugs removed, but it hadn't seen perforce since
before the 0.9.3 ISO release in october. He finally committed them on
2005/01/17, after which they unfortunately got lost in the re-org.
Benoit would generally mark bugs as fixed and never submit anything
to perforce that showed this fix, just fixing the bug locally on his
box. For fairness, it has to be said that he does commit regularly nowadays.

Enough of the conspiracy theories, let me just say I was quite pissed off
trying to get an ISO to work with an installer that had a newer version,
but for quite a while I couldn't even get my hands on it because I didn't
want to download his ISOs (limited traffic on my line, as well as squashfs)
and he didn't commit it. And I remember a quite surprised Unet saying:
Jan 15 11:09:07 <Unet> I had *absolutely* no idea what he was doing
Jan 15 11:09:20 <Unet> never had an email about 'i'm doing that'

Another memorable experience developing with benoit were his priorities,
which sometimes extremely differed from mine. There was a feature request
issued on [2] (ISO:REQUESTS, [Seth Woolley]) asking for ISO size to be cut
down where possible. - Basically by reduction of unneeded stuff on there.

To achieve this, benoit looked into compressed filesystems. He chose
squashfs as the best option and implemented this on his ISOs. Actually
a very nice thing to do. Not really neccessary to pump time into IMO,
but interesting... if it wasn't for the fact that anyone with a regular
kernel can't mount them... not all implementations of features are for
the better, it seems. This would be one example for the (from my point
of view) mis-placed priorities benoit sometimes seems to follow. There
were and are more serious bugs to fix.

or:
10:12:42 < benoit> it's like the multilib issue for instance
10:13:01 < benoit> if i was alone in the ISO Team, I would probably make the
                    next ISO with multilib

but please, don't bring up the multilib discussion again...

Okay, but enough of that. I really don't want to bad-mouth anyone, just
recount history, even though it will be tainted by my feelings.
Benoit is quite an able coder, and usually tracks bugs down to the very
source and fixes them.

When David joined the team, I thought the installer would be a nice task
to tackle and (mainly because I didn't like that work ;-) ) suggested
that he split it up into manageable parts so bugs could get fixed faster
and the thing understood by someone who hasn't meditated over it for a
month. He branched it from perforce to work on it, and even put in a notice
in the original that he was working on it and that any changes to it
may be lost and hard to carry into the new version. He has generally
been pretty close to perforce, submitting often, so we could see his
progress and work. (I remember benoit had to be bugged for months just
to submit his latest scripts - and some more weeks to find the files he
forgot to submit)

After some time, I suggested that I switch over and take some of the
installer re-org and re-code off his hands, also to speed it up because
other stuff (my scripts) needed it to be done to progress much further.
I found him to be cooperating, and also found a nice TODO file outlining
what was to be done.

Generally, I find him a very team-working person able to draw up design
drafts and good at communicating. His priorities seem to be what the team
and the ISO needs most.

So, judging by team-work, usage of perforce, communication, and
last-but-not-least setting community-guided priorities, my clear favorite
is David. I hope my little "speech" (however long, thanks for reading it)
will have convinced you of the same.

For the record: I decline my nomination for ISO Team Lead.

So much for now (I guess there hasn't been a longer ML mail for months)
  Karsten "BearPerson" Behrmann

[0] http://www.plethora.net/~seebs/faqs/hacker.html#question-3.2
[1] http://benoit.dtdm.org/iso (dead now)
[2] http://wiki.sourcemage.org/index.php?page=BenoitPapillaultWork

> On Thu, May 12, 2005 at 06:42:55PM +0200, Unet wrote:
> > Hi all,
> > 
> > I nominate Karsten Behrmann (BearPerson) for ISO TL.
> > 
> > He gave a lot of inputs in working out a new design for iso scripts and
> > architecture, and has proven very effective in planning the future works
> > in collaboration with David Kowis.
> > 
> > Cheers,
> > 
> > Unet

On Thu, 12 May 2005 09:53:37 -0700, Seth Alan Woolley wrote:

> I wish to second this nomination.  I believe Karsten has excellent 
> interoperability and is vendor-neutral, and his work on the installer 
> has always been in the open.  He makes good community-based decisions, 
> listens well to input, and acts well upon it.  Nothing against the other 
> candidates, but I like Karsten's management style the most.  He's also 
> an excellent code contributor and well-rounded.
> 
> Seth

Attachment: pgpxwi9MUUHq8.pgp
Description: PGP signature