[abcde] Re: abcde patch

[Robert Woodcock <rcw@debian.org>]

On Sun, Jan 14, 2001 at 01:51:04AM -0500, David Bergeron wrote:
> I've made some modifications to abcde and I though I would share them
> with you.

Hey, thanks!

I'll put just about all of this into 1.9.8.

> I've worked on completing the parallelism and the support for distmp3.
> 
> I'll quickly resume what I changed
> First of all I've moved $ABCDETEMPDIR to $WAVOUTPUTDIR which is 
> pretty usefull in my case since I want to store the completed MP3 on 
> an NFS server but I want to rip locally.

Ooops, I rewrote pretty much everything around WAVOUTPUTDIR and forgot
to make it work again. I don't use that feature myself. (Yes, I plan to
go through every silly little feature that's been building up since 0.2
and make sure it all still works before releasing 2.0... I just haven't
had time yet.)

> When I started playing around in the script I incorrectly thought the 
> way you were choosing the LOCATION was broken without really looking 
> into the mechanics first so I started changing it for something else. 
> What I wanted is to have something that always choose the first 
> available location in $REMOTEHOSTS instead of going round robin in 
> them (which I think is what you did using modulo if i'm not 
> mistaken).

> I guess this behavior could be changed as a config option. 

Only if both algorithms are clearly superior when you're trying to optimize
for a specific need (like speed or lack of disk space). Abcde has more than
enough subtle bugs without giving users a myriad of scheduling options. :)

I believe a poll-based scheduler such as yours would be a win for any
situation with multiple simultaneous encoding processes (whether it's
via SMP or distmp3).

> Also distmp3 can use the ":" in the hostname field as a way to 
> specify the port, so I used a space as the separator in 
> $ENCODELOCATION instead of the ":" to make it possible to specify a 
> port with hostnames in the config file.

__local__ is a possible hostname (a silly one, and a standards-breaking one,
but still...). I'll probably use something like %local%.

> I added another status file message so that the main script can wait 
> for the encoding childs to be done before cleaning up.

Remind me what was wrong with encodetrack-$TRACKNUM?
-- 
Robert Woodcock - rcw@debian.org
"The ability of the Internet to position dog excrement at the center of so
many lucrative commercial transactions is one of the most astonishing
developments of the new millennium." -- Forbes

To unsubscribe: echo unsubscribe | mail abcde-request@whimsica.lly.org