2008/1/20, Herman Robak <[EMAIL PROTECTED]>:
>   Patching Cinelerra 2.x to an acceptable level _may_ be doable this
> year.  I think the statement "I believe in Cinelerra" was a vote in
> favour of patching Cinelerra 2.x, otherwise it should have said
> "I believe in the future of Cinelerra".

So, I've seen there are quite a view packages available for cinelerra at:

http://cvs.cinelerra.org/getting_cinelerra.php

How are these packages built? I guess they are built by hand and then uploaded?

What about setting up some kind of "Build Farm" that automatically
fetches cin from the repository, uploads her to a selection of virtual
machines running different _popular_ Linux Platforms, builds packages
and puts them onto some publicly available repositories?

While I normally do not tend to recommend throwing organizational and
administrative resources at a coding problem, I feel that in this
unique situation it could make sense, and therefore I'll argue a bit
for it.

As always, these are my opinions and you are free to disagree and argue against.

As cinelerra uses mpeg and other stuff, it is somehow neglected by
official repositories of big distributions, so alternative channels
are needed. There is the Open Suse Built Service, but it prohibits
mpeg-multimedia related packages, so this is not an option either. I
am just stating this for the sake of completeness.

So, anyways, the value that a (virtual) build farm would add, is
tightening the feedback-loop between users and devs. If a user reports
a problem, the fix can be commited, and the next days fresh packages
are available, and only on "apt-get update" away. Sure, everyone could
do a svn update and a make, but I am quite confident that this is to
much of a hassle. It would be for me.

Furthermore, such a setup has the potential to nurture a more
consistent distribution of cinelerra "in the wild". That is, everyone
would be able to run the exact same version of cinelerra, which
_should_ make it simpler to reproduce bugs. Hopefully.

It makes testing easier, spotting regressions easier, spotting broken
builds will be possible, etc...

Anyways, arguing about why it should be done is not sufficient, if it
cannot be done for real.

So, what kind of resources are available for any potential cinelerra
contributors? What is necessary for this effort is some kind of
dedicated machine, that is free to be bogged down with lengthy compile
cycles on a regular basis. It needs a permanent Internet connection,
it needs to run continuously and should have a little bandwidth to
waste, and not to little RAM.  It is possible to rent such machines
quite cheaply, but "somebody" has to pay for it. Alternatively,
someone could donate hosting or a machine or both.

Hosting Mailinglists, repositories, wikis, etc. is NOT a problem, I
can do that, and I am not the only one. What I cannot do, is provide a
dedicated machine, that can burn cycles.

So, any opinions? Good Idea? Bad Idea? Necessary? Unnecessary? How to
get it done?

Cheers
-Richard

_______________________________________________
Cinelerra mailing list
[email protected]
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Reply via email to