El 09/07/2014 9:26, "Albert Cervera i Areny" <alb...@nan-tic.com> va escriure: > > 2014-07-09 5:47 GMT+02:00 Sharoon Thomas <sharoon.tho...@openlabs.co.in>: > > > > On Jul 9, 2014, at 3:41 AM, Cédric Krier <cedric.kr...@b2ck.com> wrote: > > > > > On 08 Jul 14:14, marsupo...@gmail.com wrote: > > >> On Tuesday, July 8, 2014 10:15:29 PM UTC+2, Cédric Krier wrote: > > >>> > > >>> Developers work on what ever they want, they are not forced to work on > > >>> any specific topic. So nobody can tell what will be developed in Tryton > > >>> more than what is already done (and probably published as patch). So if > > >>> you are looking for a roadmap, you are at the wrong place. > > >>> > > >> > > >> I was not talking about a roadmap of new features but a roadmap (or todo > > >> list) of features to be removed (such as multi-database support). > > > > > > Something added or something removed, it is the same. It is an > > > improvement and they fail on the same workflow. > > > > > > > That being said, I am not sure that removing multiple-database support is > > a “improvement” that has wider acceptance. > > > > I personally feel its a step backward. Multi-database support has been a > > critical advantage of tryton and we use it for a variety of reasons from having a > > demo/playground database in the same environment for users to try things > > safely along with production database to having multi-tenant systems. > > > > From the previous discussion about the topic and the available information I > > don’t see why multi-database support should be removed. If creating and > > dropping databases is the issue, we should remove that instead. > > What I've read so far, dropping multi-database support has the > following advantages: > > - (...) > - Allows improving the usage of maps/paths with WSGI (don't know the details) > > On the other hand I don't personally see many problems with removing > the feature because: > > - (...) > - Multi-tenant systems should be easy to implement with other tools, > maybe the only issue is that you'll need some more resources because > you'll have several trytond processes but I guess you need that anyway > if you want to provide a decent service
With the WSGI support and Circus (for example) have a multiprocess (to serve multiple databases) should be easy and more flexible and powerful to implement something like DB selection by subdomain.