Re: [Koha-devel] LibLime 2.4 RM Proposal

[Andrés Tarallo]

Paul POULAIN wrote:

> That problem is hard to avoid with source developped by many people.
> It can partially be solved with a solid API & API documentation (which 
> we have only partially, I agree)

It can be avoided :). We should work to enhace the API.

> > On the other hand I think that Koha will gain a lot if it get rid of 
> > the non marc database.
>
> I'm not sure at all. You must remember Koha is multi-MARC, so, a 
> "title" may be in different places. It is very CPU consuming to search 
> everytime "where is the title ?". It's better to store it definetly 
> during biblio adding process.

In the same way we "cache" some tables in C4::Context this can be done 
with "translations" between the marc field number and its meaning.  Some 
databases supprt "in memory" tables, this could speed up searches in 
tables and working in the way I'm proposing.

> > About speed: We have gained a lot with mod_perl. I encourage others 
> > to test it and make improvments.
>
> could'nt give it a try. But still expect to be able to one day...

It works, we will have in a near future our ssystem in production.  
Another enhacement to performance can be done writing some modules in C 
with bindings to PERL.

Andres