[ros-dev] when will I return Magnus Olsen aka GreatLord

gedmurphy gedmurphy at gmail.com
Thu Jun 19 09:42:18 CEST 2008


I don't have too much to say on this other than I agree with both Aleksey
and Alex.

Our current development model is a great model to get projects off the
ground, but we've passed that stage now. We need stability, code proofing
and lots and lots of testing. Magnus' commits have always been erratic and
difficult to follow, so much so that he is one of the only developers who's
patches I skip over in ros-diffs. (and I know I'm not the only person who
does this). However, I'm unsure whether it was right to single him out.
Maybe it's time to move all developers into a similar model, a model more
akin to Alex's suggestion.

I sympathize with Aleksey and the testers, who I watched many times chasing
their tails regress testing a broken tree due to untested commits. Commits
which are mostly from the same tiny minority.

Did Aleksey make the right decision? I'm not sure. But he did make a
decision, a decision which he believes is for the good of the project. I'm
happy to go along with these decisions because no one else has more of the
facts and is in a better decision to do so.

Ged.



-----Original Message-----
From: ros-dev-bounces at reactos.org [mailto:ros-dev-bounces at reactos.org] On
Behalf Of Alex Ionescu
Sent: 19 June 2008 02:38
To: ReactOS Development List
Subject: Re: [ros-dev] when will I return Magnus Olsen aka GreatLord

Man, I wonder where you guys were when I was getting shit for breaking
the kernel due to *valid* changes.... and cooperating.

I love Greatlord but, it's not the fact he's breaking stuff that I
think Aleksey is tired off... it's breaking stuff, promising to
change, and then still break stuff. Looking over commits, I don't
think anyone has broken the OS more than Greatlord except myself. I
also don't think anyone has committed more code than myself (And
probably w3seek). So yes, there does seem to be a valid proportion.
Both greatlord and I were massive regressors, but also massive
changers... that doesn't make it RIGHT though. It was probably wrong
even during my time... I can't get away by saying "ah yes, but my
changes were correct!!!!". So I will probably say that lenience was
given to me (not always though, I did get into some pretty big
fights)...

Now, because Steven Edwards made a mistake and didn't punish me enough
(I eventually stopped breaking stuff when I stopped being 18 years old
and having no SE experience, you'll notice), does it mean history have
to repeat itself? This is an age-old chidish argument "BUT HE DIDN"T
GET ARRESTED!!! WHY DID I?". If the law failed to punish someone, that
doesn't mean everyone gets to go free.

Greatlord's breakage of the trunk, although proportional to his
commits, reflects a bigger problem -- your development model SUCKS,
and you need to fix it. So I don't think Aleksey is being -unfair-...
I think he's incorrectly attacking the person who "abused" the relaxed
rules the most... what he should REALLY fix is the rules themselves,
because any other dev in Greatlord's position would've probably acted
the same. Greatlord is just being the scapegoat.

So guys, one last time.

FIX YOUR DEVELOPMENT MODEL.

Think of it:

(# of developers which you ASSUME you will lose, for having a good SE
model*)

Is that bigger or less than:

(# of developers that you have lost because they got sick of regressions +
 # number of developers that you have lost because they got sick of
being punished for doing regressions +
 # number or developers that you have lost because they got sick of
seeing their friends punished for doing regressions)?

* Not only is this an ASSUMPTION, it's a STUPID one because this is
how EVERY LARGE FOSS PROJECT WORKS.

Please, get your shit together. Use the model I gave you, as a start,
or ANYTHING else than this shit.

Best regards,
Alex Ionescu


On Wed, Jun 18, 2008 at 3:36 PM, KJK::Hyperion <hackbunny at reactos.com>
wrote:
> Aleksey Bragin ha scritto:
>> All of that was done in a deep night (~4am, european time).
>
> which happened a grand total of 5 times (all in the same day - also in
> sequence, meaning they might as well be one, from the point of view of
> diffing) since the creation of the win32k-gdi-dx branch (33435)
>
>> I must admit Magnus always listened to my rants, and he did this time
>> too, but it always happens AFTER the commit-revert spree
>
> do four reverts over the span of a week (33764, 33782, 33881, 33885)
> count as a "spree"?
>
>> And also he usually kept our agreement for a limited amount of time,
>> like this time with a branch, which lasted roughly a week.
>
> surely you mean "roughly four weeks"? (33435 to 33881 - or should we set
> the end of the "pax greatlordiana" at harmless 33598? unrelated 33632?
> third-party patch 33763? MSVCRT-related commits starting from 33764?
> where exactly did he get uppity and forget his proper place?)
>
>> It was not really an issue of trust, but it's an issue of simple
>> common sense and obeying very-very simple rule: don't *play* in an
>> official, working, fragile branch called "trunk".
>
> pretty harsh assessment of two reverts about 3 hours apart (33882 and
33886)
>
> Is there a story here apart from what the repository can tell? Is it
> just a matter of perspective? bad timing? Help me here, because it
> really, really seems to me you are alienating away a developer with
> irreplaceable experience on the grounds of a bad day you had on June,
> 7th (with a hint of handicap discrimination, for extra flavor)
>
> Apologies for turning the drama up to eleven, but it seems to me Magnus
> is getting a pretty raw deal for all he's done for us
>
> _______________________________________________
> Ros-dev mailing list
> Ros-dev at reactos.org
> http://www.reactos.org/mailman/listinfo/ros-dev
>
_______________________________________________
Ros-dev mailing list
Ros-dev at reactos.org
http://www.reactos.org/mailman/listinfo/ros-dev



More information about the Ros-dev mailing list