Re: [Gforge-devel] Architecture, LGPL

[Christian BAYLE]

Ryan T. Sammartino a écrit:

> I see a new "lgpl" directory has been checked in, and I think I know
> where that is going, and I think it's a pretty good idea.  Here is some
> random ideas:
>
> I've spent a fair bit of today looking at the code for phpgroupware, and
> I think they have a lot of good ideas that we should look to borrowing
> for a 4.0 release.  They seem to have a pretty clean architecture that
> easily allows people to hook in and create new applications (calendar,
> address book, etc).
>  
I completly agree phpgroupware has good ideas, nice plugins, nice setup.
phpgroupware is a GNU project and savannah looked at this possibility to 
use phpgroupware
Even someone for phpgroupware came to ask some questions on debian-sf 
forum.
It seems it can be a difficult task.

Roland's plugin allows also to easily hook in, do you have some idea on 
how different are
phpgroupware concepts?

> I think we could learn something from that, as well as solve the issues
> surrounding proprietary addons:
>
> A clean well architected LGPLed core with a well documented API would
> allow anyone to write addons under the license of their choosing.
> "Core" gforge would know about projects and now much else.  Everything
> else (trackers, document managers, whatever) would talk through the API
> to hook in at the appropriate places, like phpgroupware applications
> do.
>
> I also think we can learn something from phpgroupware's setup...
> basically, the first time you connect, there is a web-based setup
> routine that sets up the database, configures the connection, creates
> the equivalent of "local.inc", etc.  Pretty slick.  The only thing I had
> to do at the command-line was "createdb phpgroupware" and "createuser
> phpgroupware".
>
> Anyways, just some random thoughts while I recover from the flu and a
> badly bruised ankle.  Hint:  never turn your skates to the outside while
> attempting to block a shot.
>
>  
My idea of the use of add-on is modularity allowing e.g. to choose one 
tracker or an other
I think this is a bad choice to bet on proprietary add-ons though I 
don't deny anybody the
right or the freedom to do it.
-My reason are:
   You can't correct bug.
   You can't translate
   You can't benefit of the experience of the programmed code
   You can't enhance/modify code to fit your needs

Adding to this it may conflict with GPL, and finnaly be a licence nightmare.

Cheers

Christian