In [Forum, post 123740 by rvelices in topic 17858] improving average rate and sort order on best rated page we can read:
rvelices wrote:
It's not the right place for the discussion but I think supporting something else than mysql is not a good idea. nobody does (wp, gallery3 etc...). it's used probably by 0.5% of users and it's a pain in writing optimized queries db upgrades etc...
This is a sensitive issue but it should not be taboo. This issue is sensitive because nicolas has put many efforts in this "technical" feature. Anyway here is my opinion.
Currently Piwigo 2.2 has an experimental support for PostgreSQL and SQLite. In practice, nicolas is not contributing to dev anymore (last commit on core = 1 year ago) and consequently nobody maintains PostgreSQL/SQLite support. Some bugs are reported on SQLite/PostgreSQL and they are not fixed. The support of Pg/SQLite brings additional constraints for coding. Nearly no plugin using database is compatible with Pg/SQLite.
Only a tiny user base is interested in this feature. rvelices talks about 0.5% but I'm sure this is more 0.05% : only advanced technical users care about the database engine Piwigo uses. People interested in Piwigo for Pg/SQLite support are disappointed because the feature is buggy and incomplete. A few months ago, I've discussed with a user (very opensource oriented user) who was happy to replace MySQL by PostgreSQL but he was quite angry : in addition to several bugs encountered on basic features of Piwigo, the interesting plugins were not compatible. He concluded that it was pointless to support PostgreSQL if Piwigo was not fully compatible, including plugins.
In addition to be buggy with PostgreSQL/SQLite, the support of other databases than MySQL has brought some bugs for MySQL, and that's really not cool. Considering the number of changes in the code to make it compatible with any database, this is normal to have bugs, but we have to wonder about the relevance of being compatible with any database engine.
For sure, Piwigo 2.3 won't be able to upgrade a Piwigo 2.2 running with SQLite. That simply means that these users (if any, considering the number of bugs) will have to reinstall Piwigo from scratch.
My conclusion is that we'd better stop trying to be compatible with SQLite/PostgreSQL and focus on MySQL. The benefits would be the following:
* less pain for writing compatible queries
* less frustration for users trying to run Piwigo with PostgreSQL/SQLite
Offline
As I said before, I think we should drop support for Postgres/Sqlite
Offline
I have plained to adapt my plugins to these database but now ^^
great less pain for us
Offline
I agree. We should drop support for Postgres/SQlite...
Offline
I agree too.
Offline
Well, let's remove it !
Offline
rvelices wrote:
plg, is it official now ?
Yes. It is. We remove support for PostgreSQL/SQLite.
Let's make things simple for just now : we just remove the listbox on the install form for 2.3.0RC2. Then we can remove "pure non MySQL" files, such as include/dblayer/functions_pgsql.inc.php or install/piwigo_structure-pgsql.sql. I'll close PostgreSQL/SQLite related issues in the bugtracker.
And important thing: we can consider SQL queries will only be performed on MySQL.
Offline
Offline
Since it has many bugs anyway, I agree with droping SQLite support. But it would be nice to provide a migration plugin to make it easy to switch to mysql.
Offline
For Piwigo 2.3.0RC2, I have removed SQLite/PostgreSQL on the installation form.
I have also closed all bucktracker issues related to PostgreSQL or SQLite.
Offline
Scumpeter wrote:
Since it has many bugs anyway, I agree with droping SQLite support. But it would be nice to provide a migration plugin to make it easy to switch to mysql.
Well... SQLite support was "experimental" see [Forum, topic 15927] [Piwigo 2.1] SQLite and PostgreSQL are in experimental state, in no way you should have installed it on your live website. I understand that it's a problem for those who didn't follow this rule. I don't plan to create the migration script, but if somebody wants to create it, it would be nice to add it in Piwigo extensions.
Offline
Hi,
I'd like to volunteer to make the postgresql support revive. I have a private server with limited resources and I wanted to use only postgresql on it, so I'm motivated by this goal.
I have a few experiences in developing code that works in both MySQL and PostgreSQL, I did that for a small plugin for Dotclear (cf commit [1] where I prefered to do some stuff in PHP instead of non-standard SQL). But the current tree of piwigo can make it easy to write different queries for different engines thanks to your dblayer.
[1] https://bitbucket.org/julienw/dotclear- … af0839c499
Hi julienw,
julienw wrote:
I'd like to volunteer to make the postgresql support revive. I have a private server with limited resources and I wanted to use only postgresql on it, so I'm motivated by this goal.
I understand you can be motivated for your own installation of Piwigo, but we (Piwigo Team) have to think in a much wider context. Are you ready to test all commits on PostgreSQL? Are you ready to test all (200+) plugins/themes on PostgreSQL? For a "long time" (years)?
On one side we have seen that core developers and plugins developers are not testing on PostgreSQL and I don't want to force them to, on the other side only a tiny portion of Piwigo users care about the technical backend.
So as far as I'm concerned my opinion as not changed concerning the support of other database engines than MySQL.
So far MySQL has proven to be adapted to Piwigo needs. I see no problem with it.
Offline
Is PostgreSql so different in query syntax It would need a complete review of extensions? Knowing everything is written for MySQL, pwg_query could translate. Sorry if I'm mistaken due to my ignorance
But yes we need you to maintain the code and debugg for years, not just create and leave after months or one year. On the other hand you could be a new team member to help us on the whole core ;-)
Whatever It happens thx
Offline