Re: [Opensync-devel] OpenSync: fragmentation is harmful

[Chris F. <cd...@fo...>]

On Mon, Jan 03, 2011 at 10:43:52PM +0100, Patrick Ohly wrote:
> Is OpenSync 0.4x alive today or vaporware? Is it recommended for
> production use?
> 
> Is OpenSync 0.22 alive and supported? I know that it works for some
> users and clearly SyncEvolution is no alternative to those users unless
> it implements the same features.

I'm not claiming that OpenSync is Hot Stuff.  I think its ideas are,
but the implementation still needs work.

What I'm trying to say is that before someone tries to kill OpenSync,
maybe the replacement should be Hot Stuff.  And if the replacement is
already Hot Stuff, then OpenSync will die of its own accord.


> > The project that is most useful and supports all useful features is
> > the one that deserves to keep going.  But in the meantime, I don't
> > believe there is any harm in having both.  It's the Linux way. :-)
> 
> The harm is that in the end, we end up with no project that really has
> the desired features.

There are lots of splits in the Linux world, and they can all be seen
as harmful in some way.  But that's how the free software world tends to
work.  I know, because even my own rather niche project has a competitor
(Barry and XmBlackBerry).  In the end, the world is a richer place
because users can choose which one works better for them.  The project
that works the hardest will get the farthest.

Whether OpenSync or SyncEvolution remains is beside the point.
In the last few months I haven't had time to work on OpenSync as I
would have liked, and so I wouldn't have been able to work on
SyncEvolution either.  Until one or both are able to clearly accomplish
the desired goals of syncing, both should remain.

If SyncEvolution turns out to be the winning project, more power to you!
Honestly. :-)  But doing that organically is the more stable path.

(I don't like using the term "winning", but we're talking about
projects that should be alive or dead too, so I think it's fair.)


> Here's the catch: without support from additional developers, some
> things will never get implemented because there are areas that I myself
> cannot or do not want to support (for example, devices that I don't
> have). GPE is one example. Akonadi another. I already wrote a prototype
> backend for Akonadi, but now need a developer who is motivated to finish
> and maintain it.
> 
> If everything already works and is implemented, why should I still want
> the support of the OpenSync developers? Clearly your skills would be
> wasted when focusing exclusively on maintenance.

I don't think that "It supports Akonadi!" is the feature I'm talking
about.  That matters for end users, but for developers who have invested
in OpenSync, what matters is that the future flexibility is there,
and that the same edge cases that OpenSync has planned for are planned
for in SyncEvolution.

So in that sense, it is the SyncEvolution "engine" that must compete
with the OpenSync "engine."  If both engines are theoretically capable
of supporting the same features, then it's just a matter of picking
the side with the most momentum.  If you can prove that, you'll have
a much easier time of killing OpenSync.

But to my understanding, there are theoretical engine differences
that need to be analyzed and thought about.  Does that make sense?
And I think Graham Cobb's questions were moving in that direction.
(Perhaps more profitably than my 2 cents.) :-)


> > (and maybe even change the name, since it sounds overly linked to
> > Evolution syncing)
> 
> I considered a name change, but rejected the idea because SyncEvolution
> already had a good page ranking and was well-known under that name at
> the time. I'm still reluctant to change it - too much uses the name
> (domain, command line tool). But if a name change is considered crucial
> despite the new interpretation of it ("the missing link"), then we can
> discuss it.

Well, I *should* know better, but I'm still discounting it in my head
by habit, because OpenSync from the beginning has aimed for Everything(TM),
while the name of SyncEvolution ties it to one small platform.

This is just an impression level issue, and has nothing to do with
what SyncEvolution can actually do.  But maybe users and developers
are getting that same impression?

Truth and clarity are more important than page rank, I think.
Page rank will follow.


> So far I have plenty of users, but only one additional developer (Ove
> Kaaven, for the N900 port). I can only speculate about the reasons, but
> I guess one of them is that OpenSync is still seen as the future for
> syncing on Linux. I hesitate to describe it with such a slightly
> disparaging comparison, but it is a bit like a honeypot that attracts
> new developers who then get stuck without ever getting nearer to a
> stable release of their work.

Honesty is the best policy. :-)  You could be right.

- Chris