[Cubicweb] [RFC] folding yams, rql and logilab-database into the cubicweb repository

Christophe de Vienne christophe at unlish.com
Wed Mar 16 17:13:29 CET 2016


I am all for merging lgdb and even yams into cubicweb, but would prefer
RQL to remain a separate module.

This is because I already use it in a POC-project that aim to produce a
RQL engine on top of sqlalchemy [1].

More generally, I see little interest in having lgdb and yams outside
cubicweb (lgdb has way better competitors, and yams failed to get useful
outside cubicweb), but I think having a standalone RQL parser and solver
still makes sens if we want to facilitate experimentation for alternate

By the way, when implementing ERAlchemy, I discovered that the rql
implementation depends on implementation details of yams that I had to
reproduce. Those should be brought back into the ISchema interface.

My 2 cents,


[1] https://bitbucket.org/cdevienne/eralchemy

Le 16/03/2016 17:00, Julien Cristau a écrit :
> Hi,
> for various reasons, there is a strong coupling between cubicweb on one
> side, and rql/yams/lgdb on the other.  This makes it necessary to have
> all of these in fairly strict version sync, which causes pain each time
> this is necessary (i.e. pretty much at every major cubicweb release).
> We think it would make a lot of things easier if they all lived in the
> same source repository, with a single version number.  One possible
> downside is that this move would discourage use of rql/yams/lgdb outside
> of cubicweb.  I don't think those hypothetical uses warrant making
> cubicweb development more painful (and in the case of lgdb, I would tend
> to think nobody should base a new project on it).
> Does anyone see any other downsides to this plan, or have strong
> objections?
> Thanks,
> Julien
> _______________________________________________
> Cubicweb mailing list
> Cubicweb at lists.cubicweb.org
> https://lists.cubicweb.org/mailman/listinfo/cubicweb

Christophe de Vienne
Co-fondateur http://unli.sh

More information about the Cubicweb mailing list