Re: GSoC Proposal: Scripting Framework for Applications

[John Palmieri <johnp redhat com>]

----- "Colin Walters" <walters verbum org> wrote:

> On Thu, Apr 8, 2010 at 2:56 PM, Shaneeb Kamran <shaneebster gmail com>
> wrote:
> 
> > The main benefit of supporting multiple languages is more to the
> users of
> > the application than to its developers. Like I mentioned in my
> previous
> > mails, if a user wants to add a feature to his favorite application
> and he
> > knows only a language, say, ruby while the app requires him to write
> the
> > extension in, say, python then the user is stuck: he has to learn a
> new
> > language just to write a small piece of code.
> 
> We can't support multiple runtimes per process sanely for a variety
> of
> technical reasons; one of those is that multiple GC'd runtimes break
> toggle references useless and lead to memory leaks.  A non-technical
> argument against this is that it fragments the developer base,
> documentation, etc.
> 
> My take is that GNOME apps should pick C + one of JS,Python and move
> on with actually writing your app and fixing bugs, making it
> compelling, etc.

I would have to agree here.  As a project the proposal has merit but the issues involved make it a poor candidate for an SoC funded project.  I would much more like to see a project which adds scripting (JS or Python) to a number of applications in the GNOME stack.  Or one that documented best practices for applications exporting plugin hooks.  Those seem to be much more useful once the SoC period has ended. SoC should be about improving on existing projects, not creating whole new projects because the time allotted isn't really long enough to start something from scratch.  For that reason a project should prove itself outside of the SoC framework before we consider it for funding.  In other words a developer should be enthusiastic enough about a project to develop it regardless of funding.
 
--
John (J5) Palmieri
Software Engineer
Red Hat, Inc.