Re: [tablix-list] Changes in CVS

From: Antonio Duran <adminies.elbrocenseREMOVE@THISedu.juntaextremadura.net>
Date: Wed Dec 14 2005 - 11:01:03 CET

Hi!

Congratulations for the great work! I tested the new version yesterday
with my old files and i think it makes a good improvement. I was anxious
to get "dependent events" support. Thanks :)

Yesterday i got into porting timeblocks to the new API, but failed :(
I will be get back to it this evening, but i attach my try in case you
want to take a look.
I was first testing with only 2 consecutive events, as done in
consecutive module. Looking at the debug output i see that i do the same
calls to updater_new like you do in consecutive module, so i am a bit
puzzled.

Best regards,
Antonio.

El lun, 12-12-2005 a las 17:24 +0100, Tomaz Solc escribió:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi
>
> I would like to inform you about two quite important changes in the CVS
> version of Tablix
>
> First, I implemented a fitness value cache. My benchmarks show that it
> approximately halves the time required to find a solution. Modules that
> use fitness functions use caching automatically and don't need to be
> modified in any way. The size of the cache can be adjusted with the
> "cachesize" parameter. It is described in detail in the man page.
>
> The second change is the support for so-called dependent events. A
> module can now define an event to be dependent on another event. This
> means that the assignment of the variable resources to dependent events
> is no longer handled by the genetic algorithm. Instead the module
> provides an updater function which assigns resources to these events
> based on which resources were assigned to a specific independent event.
> This is useful for example for timetabling constraints where one or more
> events have a fixed relative position to another event (consecutive
> events, events that must be scheduled at the same time or in the same
> room). Until now these constraints (which are trivial to solve) required
> a lot of computational effort that can now be used on other constraints .
>
> Currently only the "consecutive.so" module uses this new kernel
> functionality, however some other modules could also greatly benefit
> from this change ("timeblocks.so", ...) If anyone is interested in
> writing modules for this new API, you can take a look at the source for
> "consecutive.so" and the functions in "depend.c".
>
> The modified kernel is still fully binary and source compatible with
> modules for 0.3.0.
>
> Best regards
> Tomaz Solc
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.2 (GNU/Linux)
> Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
>
> iD8DBQFDnaRXsAlAlRhL9q8RAr3yAKC6sFt+VV1WncPKd47CIyb67ol3agCfZvzN
> F9XLDiBs0YAJZ1++oXsFImY=
> =kbbe
> -----END PGP SIGNATURE-----
>

Received on Wed Dec 14 11:03:07 2005

This archive was generated by hypermail 2.1.8 : Thu Dec 15 2005 - 06:27:27 CET