Re: [tablix-list] timeblocks and timeblocks_sameday modules

From: Antonio Duran <adminies.elbrocenseREMOVE@THISedu.juntaextremadura.net>
Date: Wed Nov 02 2005 - 09:24:21 CET

Hi

El lun, 31-10-2005 a las 19:57 +0100, Tomaz Solc escribió:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi
>
> I have a few more comments on your timeblocks module.
>
> If I understand the fitness function correctly it doesn't recognize all
> valid combinations of events:
>
> For example, if we have two block of three events it correctly
> recognizes for example these
>
> first block: 0 1 2 second block: 3 4 5
> first block: 1 0 2 second block: 5 4 3
> etc.
>
> but doesn't recognize these which I think are also valid
>
> first block: 3 0 5 second block: 4 1 2
> first block: 4 1 0 second block: 2 5 3

I don't think so, but i can't be plain wrong :) I think i wrote it from
your consecutive2 without even understanding all the logic inside it, so
it could be.

>
> Also I think the module doesn't behave properly if number of repeats
> isn't dividable by the number of events in a block.

Yes, i forgot to add a check for that. But when i realized, i thought
that i should better try to rewrite it in a more general way, allowing
this kind of restrictions:
<restriction type="periods-per-block">3 2 2 1</peridos>

With a variable number of periods.

This way, you could simplify your sample2.xml example.

Reading the blog, maybe it could be better to wait.

>
>
> I'll add your modules to the CVS as soon as I'll get around to writing
> automated test cases for them.
>
> By the way: I'm planning some changes to the kernel that would greatly
> simplify writing and improve performance of modules like
> "consecutive.so" and "timeblock.so". See section on "dependand tuples"
> at http://www.tablix.org/wiki/wiki.pl?Tablix_0.4

Well, this sounds great, becasuse it's true it seems hard for tablix to
do the consecutive thing.

I looked at the genetic alg recently, and thought about something
similar to what you say in the blog, although i haven't done any
research yet: we would divide the genetic alg in parts, and each module
would ask for some of them.

I think i am going to start testing with the set of my timetibles that
don't need consecutive hours, while you think about dependant tuples.
In fact, only some of my timetables need that feature, it was just
casual that i started with them.

Best regards,
Antonio.

>
> Best regards
> Tomaz Solc
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.2 (GNU/Linux)
> Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
>
> iD8DBQFDZmkrsAlAlRhL9q8RAjw8AKDkdCv6KahX55SFFp9aVPaOw5MqhgCg4hJ8
> HVRTyk2SmMxZzREL8aPAEBI=
> =Oua1
> -----END PGP SIGNATURE-----
>
Received on Wed Nov 02 09:25:49 2005

This archive was generated by hypermail 2.1.8 : Thu Nov 03 2005 - 06:27:41 CET