[maemo-developers] Community widgets for Fremantle

From: Murray Cumming murrayc at murrayc.com
Date: Thu Oct 1 14:46:24 EEST 2009
On Thu, 2009-10-01 at 11:11 +0100, Graham Cobb wrote:
> On Thursday 01 October 2009 10:16:56 Murray Cumming wrote:
> > I don't see any need whatsoever to fork hildon. It's just an invitation
> > to confusion and nasty problems such as this.
> >
> > If someone wants to create a reusable set of widgets, they should just
> > create a new library. It's not difficult. What's the downside?
> Quick summary of many emails: there are two options, both have 
> disadvantages...
> 1) New library: clean solution but problems are mainly that the widgets will 
> never be included in Hildon, may not be consistent with HIldon (particularly 
> as changes get made), the community will probably lose interest in maintaing 
> them, they will not be in the standard SDK or documentation, many developers 
> will never know they exist.  And it is yet another widget library on top of 
> GTK: Maemo already has one (Hildon), do we really need two?

These disadvantages are nothing compared to the disadvantages of
forking, or even of adding (as yet unwritten) new third-party code
directly to a stable library. And it assumes that trying to put widgets
directly into hildon will magically improve them, maybe by therefore
forcing hildon developers to work them. But that won't work - they will
just rot there and eventually be removed.

Just do what's normal. Create a library. If/When things are good enough
then they can go into regular hildon. Writing good APIs is very
difficult - it takes time and a few iterations.

> 2) Put in Hildon: great long term solution (addresses all the disadvantages 
> above).  Main disadvantage is that Nokia (not the Hildon development team) 
> control Hildon updates -- it is not posible for the community to provide a 
> package which updates Hildon on the device except as part of an official 
> Nokia SSU.  Nokia control not only the timing of such a release but may (for 
> reasonable business reasons) even decide not to ship an updated Hildon in an 
> SSU because the risk of introducing a bug is larger than the benefit of the 
> new widget.  Also, users may decide not to install the update for similar 
> risk concerns (maybe less likely in this age of frequent security updates).  
> So, a Hildon fork might happen if applications really want to use the new 
> functionality.
> There is no good answer.  I believe the only long term fix is for Nokia to 
> (effectively) fork all the libraries they use and put them in a completely 
> separate directory (/usr/nokia/lib), and link their own applications to use 
> that directory (not forgetting to release source packages for their forked 
> libraries).  They would then allow the community/user to install and update 
> any libraries in the standard directories, for non-Nokia apps to use.  Of 
> course, any developers who trust Nokia more than the community (their choice) 
> could choose to use the Nokia libraries.
> Graham
> _______________________________________________
> maemo-developers mailing list
> maemo-developers at maemo.org
> https://lists.maemo.org/mailman/listinfo/maemo-developers
murrayc at murrayc.com

More information about the maemo-developers mailing list