Re: Reboot: Strategic goals for GNOME

[Philip Van Hoof <pvanhoof gnome org>]

On Wed, 2010-03-03 at 18:46 -0500, Jud Craft wrote:
> On Wed, Mar 3, 2010 at 5:35 AM, Andrew Savory wrote:
> > Focusing in on one area that I can talk about: Qt is perceived by
> > some to be stronger from a business perspective due to the 'more
> > complete' offering: extensive documentation and an SDK.

Correct.

I have seen some people saying that Qt was picked over GTK+ by most
recent developments of the new Maemo platform because "Nokia couldn't
buy GTK+, but they could buy Qt".

I think that's quite incorrect. AFAICT they had a big problem finding
competent developers, and big problems getting GTK+ to become more
innovative for mobile use cases. Mobile is changing way faster than GTK+
is, and that's a problem.

As I mentioned before in the earlier thread, I think it's a self
inflicted problem.

Putting the blame on Qt being buyable is being in la la land.

> > Perhaps more focus on and promotion of GNOME's developer tools/sdk
> > offerings would be a useful meta-goal for the coming year?

Yes

/me whispers Anjuta

> > Somehow enunciating the proposition that you don't need to be an
> > alpha-dog developer to get engaged with GTK etc.

I agree

/me whispers Vala

> > For example, I only recently found out about Anjuta: it's presumably
> > a fairly important tool for people developing using GNOME
> > technologies, but look at the results at
> > http://www.google.com/search?q=anjuta&as_sitesearch=www.gnome.org
> > (Yes, I know there's a ton of stuff at library.gnome.org, I'm being
> > devil's advocate here ...)

> Looking at Anjuta, I have no idea if it's a great resource to start
> GTK programming with or not.  You say yourself "presumably", and
> that's the greatest nail in the coffin - you're obviously involved in
> GNOME development and you have *no* idea, you're barely familiar with
> it either. Otherwise I'm pretty sure you'd use words a little less weasely
> about it.

GNOME developers (not to use "we") don't dogfood Anjuta enough. I know
that some developers, like myself, use it on a daily basis. I have also
been filing a lot of mini bugs about small problems about it.

I must say that the Anjuta team are very responsive compared to other
GNOME components when it comes to addressing and fixing those.

So I think it's ready for some serious dogfooding.

> You don't have to be an alpha dog to realize that GNOME has no blessed
> development workflow.

For a project like Tracker it comes down to:

git pull
git branch newfeature
git checkout newfeature
Change src/Makefile.am
Possibly change configure.ac
touch src/newsourcefile.c
Create src/newsourcefile.c
git add src/newsourcefile.c
git commit -a
git push origin newfeature
/msg #project Hi! I just implemented newfeature in branch newfeature
/msg #project Sure, thanks for review, I'll push to master
git rebase master -i
git push origin newfeature:master
git push origin :newfeature


> Currently I don't program in GNOME/GTK.  I have no idea how people
> actually *are*, since GNOME has (almost by intention) no approved
> development environment (a liveCD full of every Linux development tool
> known does not count).  I assume most of them are probably just
> writing their code by hand in Vi and passing esoteric arguments to
> GCC.  Serious, I have no idea how real GNOME developers program in
> GNOME - and my guesses aren't flattering.  [If it's anywhere near my
> guess, then no, I won't be programming in GNOME anytime soon.  And I
> can use Vi just fine, and GCC with some effort.]

With Anjuta's gnome-build integration you can do most of the build
environment changes using the popup menus: it'll adapt your Makefile.am
without completely rewriting it (being afraid of that is why, I think,
most people change Makefile.am manually).


> In other words, I think I have to be an alpha-dog developer, and
> nothing I've seen convinced me otherwise.  There's just too much crud
> to wade through, pulling together API references, documents on GUI
> design, etc. (I still have no idea what GtkBuilder is, and if I should
> even still try making a GUI in Glade or not.  I hope you guys really
> don't write the XML by hand now.)

No, Glade-3, GtkBuilder or the integration in Anjuta

> And any tutorial that starts with describing how to manage and link
> my object files on the command line isn't going to convince me
> otherwise.

I think there are some tutorials on subjects like this already, but
you're right that it all isn't very coherent.

I recall that there was at some point a book written and published about
GTK+ development. I think it's quite outdated now. Perhaps a team should
step in to bring the sources of that book up to date, and get it
republished?



Cheers,


Philip

-- 
Philip Van Hoof, freelance software developer
home: me at pvanhoof dot be 
gnome: pvanhoof at gnome dot org 
http://pvanhoof.be/blog
http://codeminded.be