This goes along with my other email about modularizing authentication. My feeling is that most of koha is sort of embedded, like auth the storage should be modularized, like this there should be a wrapper with default methods that do something a default way, one of those would be search. the mysql search would use mysql features (ie: full text search), a postgres search would do whatever they do. basically you'd have custom queries for each DBM since each method that does the querying would be DBM specific. Only the parent class would hold static variables and things that would be consistant across all the DBM's. If done this way storage is trivial, one could store everything in flat text files if they wanted!. I must emphasize that this is the *correct* way to program things like this, and if we (my university) decides to use Koha I will certainly help wherever i can. Joshua Brindle UNIX Administrator Southern Nazarene University
Chris Cormack 01/16/03 04:10AM >>> On Thu, Jan 16, 2003 at 11:02:43AM +0100, paul POULAIN said: Chris Cormack a ?crit:
Hi There
Just to throw in my 2 cents worth, i dont see using mysql fulltext searching as ruling out db independence. It just forces us to code more carefully, perhaps setting a value in
/etc/koha.conf usefulltext=yes or the like. Which we can then use to
decide whether to make a standard like search, or to use a fulltext index.
BUT : the standard like search is UNUSEABLE on a 50 000+ biblio DB, as every search means a full parsing of the table. If you add the same thing for the thesaurus table, which is widely used in marc cataloguing, and some other place where full text indexing will be usefull...
Hi Again paul :) Well maybe not 50,000 .. but 500,000+ ... its working ok for HLT's 78709 biblios. But I take your point, and agree wholeheartedly the current search code badly needs a rewrite and lots of optimisation. And im sure HLT wouldnt say no to any speed increase :-) I think what i was trying to say (bear in mind its getting late here, and my mind is probably muddled :-)) was that we should be able to provide the option for other people to use a database of their choice. Without restricting those who want to from using mysql fulltext indexing. On the topic of fulltext searching we had a discussion about it on irc today. I cant currently get to the irc logs but when i can, ill mark the discussion and post the url to the devel list. Chris -- Chris Cormack Programmer 025 500 789 Katipo Communications Ltd chris@katipo.co.nz www.katipo.co.nz _______________________________________________ Koha mailing list Koha@lists.katipo.co.nz http://lists.katipo.co.nz/mailman/listinfo/koha
Joshua Brindle a écrit:
This goes along with my other email about modularizing authentication. My feeling is that most of koha is sort of embedded, like auth the storage should be modularized, like this there should be a wrapper with default methods that do something a default way, one of those would be search. the mysql search would use mysql features (ie: full text search), a postgres search would do whatever they do. basically you'd have custom queries for each DBM since each method that does the querying would be DBM specific. Only the parent class would hold static variables and things that would be consistant across all the DBM's. If done this way storage is trivial, one could store everything in flat text files if they wanted!. I must emphasize that this is the *correct* way to program things like this, and if we (my university) decides to use Koha I will certainly help wherever i can.
That's it, i have understood what you meaned yesterday, with this mail. There is an important thing you must know : koha IS NOT OBJECT ORIENTED developped. We spoke about this a few months agos, and agreed : * it was a shame * it was to hard to change this at the moment. In fact, most scripts work like this : * decode the cgi parameters * open template * depending from the parameters do something * retrieve data for next screen * send html and there is no object here. It's standard programming. The only object we have is the C4::Context, which has been developped by Andrews a few months ago. Note however, that all the sql code is in C4 packages, so, it's a kind of "modularization" :-) -- Paul POULAIN Consultant indépendant en logiciels libres responsable francophone de koha (SIGB libre http://www.koha-fr.org)
Hi All This has become a really interesting and useful discussion. With lots of great points to consider in the upcoming rewrite/refactoring of Koha code. (After 2.0 .. one can of worms at a time :-)) It also serves as a great illustration of the principle beauty of free software. That is, since we all have access to the source code we are able to have an informed and frank discussion about what we do and dont like about it. And then are able to work together to fix the things we dont like. So in this vein, I encourage all the participants in the discussion and any others interested to write up a page on the wiki. http://www.saas.nsw.edu.au/wiki/index.php?page=KohaProject Collating the ideas that we'd like to see implemented in the rewrite. So that we dont lose any of these ideas, and hopefully a bit of a roadmap for 2.1 will start to appear. Chris -- Chris Cormack Programmer 025 500 789 Katipo Communications Ltd chris@katipo.co.nz www.katipo.co.nz
On Fri, 17 Jan 2003, Chris Cormack wrote:
Hi All
This has become a really interesting and useful discussion. With lots of great points to consider in the upcoming rewrite/refactoring of Koha code.
(After 2.0 .. one can of worms at a time :-))
It also serves as a great illustration of the principle beauty of free software. That is, since we all have access to the source code we are able to have an informed and frank discussion about what we do and dont like about it. And then are able to work together to fix the things we dont like.
I'd really like to second this. I think this is really the key win that free software provides. I don't think you'd ever see a discssion like his happen at a (proprietary vendor's) Users Group meeting. On the other hand, this level of freedom comes with a price, we need everyone to get involved in these discussions to help push Koha along. Even if you aren't a developer (and don't want to be), your input is certainly wanted in terms of: what makes a good serials package, what should we do for acquisitions/circulation, what about different kinds of searching or ranking, what kinds of features should be added or enhanced ...
So in this vein, I encourage all the participants in the discussion and any others interested to write up a page on the wiki. http://www.saas.nsw.edu.au/wiki/index.php?page=KohaProject Collating the ideas that we'd like to see implemented in the rewrite. So that we dont lose any of these ideas, and hopefully a bit of a roadmap for 2.1 will start to appear.
If you don't feel comfortable working on a wiki (they're not that hard though), please feel free to talk here or on the koha2010 mailing list[1]. (If you are comfortable using wiki's, please feel free to wiki-ize conversations like this so that they don't get lost.) -pate [1] koha2010 is a mailing list set up to discuss the strategic direction of Koha. The discussion there should be non-technical (from a computing perspective). You can subscribe by sending email to koha2010-subscribe@kohalabs.com
Chris
-- Chris Cormack Programmer 025 500 789 Katipo Communications Ltd chris@katipo.co.nz www.katipo.co.nz _______________________________________________ Koha mailing list Koha@lists.katipo.co.nz http://lists.katipo.co.nz/mailman/listinfo/koha
Chris Cormack a écrit:
Hi All
This has become a really interesting and useful discussion. With lots of great points to consider in the upcoming rewrite/refactoring of Koha code.
(After 2.0 .. one can of worms at a time :-))
It also serves as a great illustration of the principle beauty of free software. That is, since we all have access to the source code we are able to have an informed and frank discussion about what we do and dont like about it. And then are able to work together to fix the things we dont like.
So in this vein, I encourage all the participants in the discussion and any others interested to write up a page on the wiki. http://www.saas.nsw.edu.au/wiki/index.php?page=KohaProject Collating the ideas that we'd like to see implemented in the rewrite. So that we dont lose any of these ideas, and hopefully a bit of a roadmap for 2.1 will start to appear.
Chris
<clap><clap> Standing ovation for this mail ! I 10000% agree, and would have written it :-) -- Paul POULAIN Consultant indépendant en logiciels libres responsable francophone de koha (SIGB libre http://www.koha-fr.org)
participants (4)
-
Chris Cormack -
Joshua Brindle -
Pat Eyler -
paul POULAIN