Re: GNOME Javascript questions and a bit of presentation

[Alexandre Mazari <amazari igalia com>]

Heya Edu,
Le samedi 24 septembre 2011 à 11:54 +0200, Eduardo Lima Mitev a écrit :
> Hi,

> 1) Is there motivation/interest in the GNOME javascript community to
> implement CommonJS [2] API specs? CommonJS intends to "standardize" a
> set of common APIs to be implemented by the different JS
> engines/platforms to improve code reuse and harmony among them. It has
> gained in popularity and today almost every JS developer do try to use
> CommonJS APIs as much as possible.
> 
I'd +1 such a move. Usin common idioms to javascript developers and
taking advantage of the rich eco-system around CommonJS would be a huge
gain.


> I would be specially interested in the reasons behind negative answers
> to this question.
> 
> Just as an exercise/experiment, I recently started writing a wrapper
> to gjs that would provide CommonJS support. Code is at [4] and only
> 'Modules' (require) is implemented at the moment. I know this is not
> the way to go in the long term but it could perhaps persuade the right
> people to push for this alignment if the experiment succeeds (meaning
> people use the wrapper instead of plain gjs).
> 

Great.
The modules importation spec looks more robust and realife-tested than
the current seed/gjs importer js API.

 Did not look at the code but I am sure there is no 'let' in there,
right ? :)

> 2) Do we have an agreement for the long term, about whether gjs or
> seed will finally be "the" engine? (or none and instead we pick
> python/pygobject?) I suppose we are not willing to maintain and push
> for both. My guess is that seed would eventually be the chosen one as
> it has to be already there anyway for WebkitGTK+. I did read Colin
> Waters' thread [3] "gjs/seed" but I would like to get an explicit
> confirmation (if there is one).
> 

JavascriptCore still lakes some standardization-pending stuff (let,
tupple deconstruction...). The project Harmony making progress, that
should appear in WebKit "soonish".
Seed seems indeed the natural choice, concidering epiphany, yelp,
empathy using webkit. But Gjs had a lot more active development in the
latest year(s), making it maybe a more proven solution.

We could:
- port gjs to current javascript standard
- abstract gjs from the javascript engine, an do a jscore backend. Risky
- make a bridge exposing mozilla API over jscore.
- rock on seed


> 3) What is the current state and the plan/roadmap to provide binding
> documentation of GNOME libraries? I know this was the focus of last
> G-I hackfest in Berlin and maybe there is already a "favorite"
> solution to the problem.

There is a great API reference at
http://www.roojs.org/index.php/projects/gnome/introspection-docs.html

> 
> 4) Do we have any clue about the right developer tools we need for
> writing GNOME apps in binding languages? Can we talk about a future
> GNOME SDK or something of the like?
> 
We should. Our current developer experience  might feel painful for
casual developers. I am not sure js is the right answer as a general
purpose language of Gnome, but may be right for ui logic.

> 5) Is there other specific message our community would like to state
> to that audience, or specific feedback to obtain from them? I would be
> happy to proxy these messages too :).
> 
> I do thank you all for taking a few minutes and help me communicate
> GNOME in this other forum.
> 

Thanks for all your ideas!