menu editing griefs: How collaboration did not work. But it should work!

[Christian Neumair <chris gnome-de org>]

Dear developer community.

I'm a bit disappointed by how the GNOME menu editing journey went.

In the very beginning, a menu editing API on top of the XDG menu spec
was developed for GNOME, by Mark and Frederic, which was and still is a
very clean and good API.

I stepped up and implemented a C menu editor called gnome-menu-editor,
entirely based on proposals Calum made [1] that were highly appreciated
by the community. I invested lots of time into making it behave
according to Calum's suggestions, it was really a fruitful task and we
had a frequent exchange of thoughts and source code. It looked like we'd
soon have a usable reference implementation.

Suddently, shortly before the end of a development cycle, Mark came up
with another menu editor [2], called gmenu-simple-editor, which was also
based on Calum's proposal, but as far as I know Mark didn't really
arrange detailed usability aspects with Calum. I wasn't sure why he came
up with it, and supposed that he might have some different innovative
ideas in mind, or maybe just preferred Python, but then again we could
have used Alacarte as a base.  Because Mark told me (in private) that my
GUI wasn't simple enough, I shifted the advanced stuff into context
menus. At some point the GUI exactly matched gmenu-simple-editor except
for more features, which were only accessible through the context menu,
shortcuts and DND, but now the UI didn't entirely match Calum's proposal
anymore.

>From the followup discussion to [2] ([3a], [3b]), it was very obvious
that I were greedy for collaboration and exclusively interested in a
great joined effort that would satisfy most of our users. I was even
willing to blow my whole codebase away if it wouldn't match quality
requirements. Make sure that you read these posts before presuming
defamation.

Unfortunately, time went by and I received no development feedback, just
some user criticism, and no offers for collaboration. I got less and
less enthusiastic, and not really integrated into platform/desktop
development, although I tried to contribute as much as I could to
gnome-menus improvements.

gmenu-simple-editor was picked as GNOME menu editor, although it wasn't
as feature-rich as gnome-menu-editor. Nevertheless, no discussion was
raised at the ddl which menu editor should be picked, possibly because I
failed to bring the issue up again - so it was also my fault, but maybe
it should also have been brought up by other interested parties. I was
very reluctant to initiate collaboration discussion again, because I
made a blanket-check offer (read: [3a]), and it might have looked as if
personal motives played a major role (i.e. pushing "my" implementation).

gmenu-simple-editor still was exposed to many user rants.
gnome-menu-editor was quite popular these days and is still on place 4
of the gnomefiles.org download statistics [4]. At this time I was quite
pissed off and stopped gnome-menu-editor development, having spent many
hours on it, and was very angry that my concerns were ignored. People
had to install a gnome-* package, where something not included in the
platform was installed, and a gmenu-* binary was shipped with a gnome-*
package.

Roughly at the same time, Travis Watkins wanted to have a feature-rich
menu menu editor [5], totally not modelled after Calum's proposal. It
was called smeg (and later renamed to alacarte), but more and more
converged to Calum's ideas. It was and is a good piece of software, but
it turned out to be the third implementation of the same concept with
some minor tweaks.

Short story: We had three implementations, and none was really perfect,
until Alacarte became more mature and Calumistic (it's almost perfect
now), and just became the default menu editor.

But because there was no communication channel between the developing
parties established, we just wasted many valuable hours.

This (possibly one-sided) report shows us how collaboration does not
work (lack of communication), and remind us that parallel development of
core components with very similar targets in mind has to be watched
carefully and as heavy redundance occurs, this should be brought up on
the relevant development lists.

I just want to make sure that something like this doesn't happen again,
maybe for Alex Larsson's great ideas on the next generation of the GNOME
VFS - so you may want to watch out for wasted efforts and point them
out/bring them up as you encounter them.

[1] http://www.gnome.org/~calum/usability/specs/menu-edit/
[2]
http://mail.gnome.org/archives/desktop-devel-list/2005-April/msg00069.html
[3a]
http://mail.gnome.org/archives/desktop-devel-list/2005-April/msg00075.html
[3b]
http://mail.gnome.org/archives/desktop-devel-list/2005-April/msg00115.html
[4] http://www.gnomefiles.org/app.php?soft_id=867
[5] http://www.realistanew.com/2005/03/18/gnome-menu-editor/

-- 
Christian Neumair <chris gnome-de org>