[compiz] Re: BlurFX for compiz

[Colin Guthrie]

Mike Dransfield wrote:
> Speaking as a developer, not a compiz developer... I think that using
> IRC for technical discussions is totally wrong.  Writing software
> requires a lot of formality, discussion about changing the core should
> be done in a public and recorded way so that people in the future can
> see why and how changes were made.
> 
> I think that the informal attitude has got us in the state we are today,
> patches applied all over the place without formal documentation.  If I
> am correct, the original xinerama changes were made with the
> understanding that they would be removed once David had finished the
> correct changes.  Now the argument is that it is too hard to undo those
> changes.  There are actually some good features that were put into
> compiz-quinnstorm which were not offered upstream because 'David would
> not accept them anyway'.
> 
> Making it easy for people to contribute is great, but what is wrong with
> discussing things on this list before changing core items?

I agree in principle, but it has to be said that one of the driving
factors of compiz-quinnstorm => beryl was lack of response from the
offical compiz crew (basically David (*ducks*) ;))

One of the great things about IRC is the ability to meet with and talk
quickly about changes and then do them and see them quickly.

Beryl as always been about RAD.

It's a credit to David that sooooo many people have become so
enthusiastic so quickly about compiz. It's just a shame that this wasn't
 managed slightly better (not a fault of David per se as you cannot
predict the will of the community!) to make the whole process more open.

Yes the Xinerama changes are/should be temporary until the "right" way
has been implemented and yes it probably will be a pain to strip it out
again, but I'm sure it will get done.

And it goes both ways. You said "There are actually some good features
that were put into compiz-quinnstorm which were not offered upstream
because 'David would not accept them anyway'". This is true, but then it
 is also up to David and others with commit access to compiz to take the
parts they like out and modify them to suit their needs. It's not as if
anything was done behind closed doors.

Now it seems that since the fork happened, that a *lot* of work has been
done on compiz which is really great :) I'm not sure if it's just
good/bad timing or if it's a matter of healthy competition, either way
I'm happy for both sides.

For me, I needed compiz-quinnstorm as I predominantly work in a dual
head evironment. I'm looking forward to when dual head is done right in
compiz, but for now I *need* patches to get a working DE and for me a
branch/fork is better.

> I think that the quinnstorm branch was a great idea and the forums help
> with support and training developers.  Things went too far and now it
> seems like the beryl developers are actively looking to break
> compatibility so they can differentiate themselves from compiz and be
> included in the distros.

The name change is indeed important for distro inclusion. I work as a
packager on Mandriva and it was a royal PITA trying to get compiz and
compiz-quinnstorm usuable at the same time (or rather installed at the
same time)

For me a name change is the right thing to do.

As for whitespace changes and function name case changes I cannot really
comment as I've not actually looked at the source since I did some very
early patching on compiz (and created one of the few patches that David
did accept :)), but one thing that struck me when working on the code
back then was a total lack of whitespace control. It was very hard to
know how to indent code blocks and how to name variables.

Again as my recent code eperience is limited this may be out of turn,
but I would strongly favour an effort by someone to go through the code
of both compiz and beryl and standardise indentation and case to an
agreed (by both communities) standard - this will reduce such problem.

I was involved in many discussions on IRC and one of the aims of beryl
was to do such an effort on the code such that developers would have a
clear understanding of how to format the code. This is the reason it
changed, but I feel that an effort from both sides is needed to make
this truly worthwhile.

> If I am not mistaken, the compiz-quinn cvs is now offline and any
> history is lost. 

Nope: http://metascape.afraid.org:13666/cgi-bin/viewcvs.cgi/?root=cvs

> I want to avoid all this political nonsense which is
> only hampering development which is why I will probably host my own
> repository.  I will try to get ifdef patches put into the beryl plugins
> so that they can work without modification.  A nicer solution would be
> to patch the parts of beryl core that cause the biggest headaches for
> compatability (BERYL_ABI_INFO I am looking at you).

Can't comment on this as I'm not a developer but it seems sensible
taking your comments at face value.

> Maybe a beryl developer can comment on my 4 points, I would really like
> to know more.

Like I say not a developer, more of an interested onlooker :)

Col.