[Bf-committers] Re: Google Summer of Code Idea

[Peter Schlaile]

Hi,

regarding good sequencer improvement ideas, please take a look here:
http://wiki.blender.org/index.php/Requests/Sequencer

regarding sql-database: why do you need one here?

regarding proxy-servers: please wait, until the new IO-plugin system is
ready, since then it is a seperate plugin. I don't think, that blender
should automatically fork a seperate server and do fancy things, most
people do not need. Should be done on request in a seperate plugin.

The data path should be

frameserver -> framserver-input-plugin -> plugin-interface -> blender

BTW: I honestly can't see, how low-res proxies can have any real benefit
from a seperate server process (like Alexander pointed out, it boils down
to rendering to some lowres cache files and using them instead of the main
files. So why have so much hassle with a seperate server?).

Input IO plugins should have the (optional) ability to fork into seperate
processes for speed and stability reasons though. (Currently the sequencer
can't benefit from SMP...) This should be possible IMHO for _any_ plugin
of blender. (I do know, that this is _not_ a good idea for material
plugins, but the option would be nice for testing purposes here...)

But running the whole input over a seperate server (bottleneck) is bad
design(tm).

Just my 2 cents.

Greetings,
Peter