[Cubicweb] extraction of legacyui in progress
aurelien.campeas at logilab.fr
Mon Feb 11 11:38:17 CET 2013
Le 11/02/2013 01:10, Nicolas Chauvat a écrit :
> Hi List,
> If you read http://www.cubicweb.org/blogentry/2709670 you know that a
> dramatic downsizing of cw/web/views is planned for cw 3.16. This
> involves extracting a lot of code to the new cube legacyui. This
> cube will therefore implement the "one ui to rule them all" motto that
> should not be part of the CubicWeb framework (thanks orbui for
> demonstrating it, sorry for not seeing the obvious).
> Here is a preview of my work in progress for those interested:
I hope you're aware that at least Rémi did work on the cw/legacyui
split recently. And that late cw 3.16 patches (from after the sprint)
probably touched more than 70% of all the views modules (due to the
move of uicfg to views).
> The first line is a changeset that begins extracting the code from
> cw/web/views and provides backward compatibility for the cubes that
> currently import from this place (and there are many).
But only basetemplates, breadcrumbs and the primary views are
moved. This is just a beginning isn't it ?
> The second and third lines are links to my two repositories for
> legacyui (the good old ui using html4) and squareui (the same ui
> choices, but with bootstraped html5).
This does not look good: we add a lot of brittle boilerplate and there
is an untold amound of duplicated code here (I'd guess the code is 98%
like the one in legacyui).
> To test the backward compat, create an instance and
> add_cube('legacyui'). It should work as it would with cw 3.16.
> You can add_cube('squareui') instead, but most parts of the ui have
> not been ported yet and as you will quickly realize, the basic cubes
> also need to be ported, and this will take time.
> If you want to preview a cube ported (in part) to squareui, try
I'm quite concerned with this way of doing things that:
* we have a fork of the ui in separate cubes
* we will have to maintain forever (for a long time at least) an
internal fork of these in our standard library of cubes
We should probably try something a bit different.
More information about the Cubicweb