qreator-discuss team mailing list archive

[David Planella <david.planella@xxxxxxxxxx>]

Al 24/11/12 17:04, En/na Stefan Schwarzburg ha escrit:
> Hi David,
> 
> today I spent some time waiting with a smartphone in my hand, and used
> that time to research PySide and QML.
> 

Hey Stefan,

Thanks for your time looking into this. Let me try to reply inline.

> I have to say, that I'm less than impressed. Actually, PySide seems to
> be even more verbose and convoluted than PyGObject, and I already
> thought that this was at the edge of being acceptable.
> 

I can't reply on this one, as I haven't played with it much yet, but
unless it's extremely convoluted to write the code, I'm not personally
much concerned about it if the end result is going to be a significant
improvement.

> And QML, while a step in the right direction from the Glade ui files, is
> very Javascript heavy. It seems as if designing is not really possible
> without JavaScript.

In my little experience with QML, my understanding is that:

- You can design apps with QML alone
- For more complex apps, you might need to add Javascript
- For even more complex apps, you'll need to resort to C++ (or in our
case, Python)

But, of course, this will have to be put to the test to see if it really
applies or I'm just talking nonsense :)

> From the perspective of a C++ developer, QML is certainly a step in the
> right direction. Developers and designers now have to write much less
> C++ code, and I understand that this is a good thing. Now, GUI heavy
> applications can be written with Qt, but completely without C++.
> But from the perspective of a Python developer, JavaScript is not a step
> in the right direction (at least not for me).
> 

Right. To be clear, it's my intention to stay on Python too, and not go
into the C++ or JavaScript realm. Now, if we have to write a small
amount of JavaScript in addition, and for UI purposes only, I don't see
it much of an issue. Assuming that our logic and core code is in Python.

> I think that if we really set our minds to developing Qreator with
> JavaScript, we should really use HTML5+CSS3. This would allow us to
> completely ignore the desktop, be open for every OS, and let the
> integration be done by Ubuntu-webapps or similar.
> However, I don't like the idea of using JavaScript. Actually, since this
> is a project that I like to do in my spare time, I would prefer to stick
> to Python.
> 

I agree with the point of going to HTML5, but not because of a potential
switch to JavaScript. If we were to go to HTML5, the two main 'pro'
points would be:

- It is a web-wide standard
- It would allow us to have the same code running on the web and on the
desktop

However, if we were to do that, I'd rather use the approach of embedding
the UI into a webkit view and have the backend still be Python (with
perhaps a small amount of JavaScript if CSS3 would not allow us to to
everything we want to do). There is some example code in the Quickly
HTML5 template at https://code.launchpad.net/quickly-community-templates

Having said that, I don't think HTML5 is ready for our purposes yet, as
I'm still primarily considering Qreator to be a client app, not a web
app. The spanner(s) in the works for me are:

- No native widgets
- No Unity integration
- No internationalization support

> So my suggestion would be to focus (after that release) on the
> separation of GUI and background-logic and then we can explore Qt, but I
> would then also like to explore kivy. The kivy language is very much
> like QML, but integrates python in stead of JavaScript, and kivy itself
> is the most pythonic GUI-toolkit I have seen so far.

I fully agree. However, although it looks pretty awesome, I'm still not
convinced by kivy.

For the purposes of exploring our choices, I've created
https://wiki.ubuntu.com/Qreator/GetInvolved/Brainstorm/GuiToolkitChoice

If you'd like to advocate for kivy, would you mind filling in the kivy
part, so that when we make the actual decision we can base it on the
facts from our research?

Thanks.

> I'm aware that kivy has many disadvantages, especially the lack of
> support from Unity specific tools (dbus-menu, HUD, Gtk-themes, ...) but
> I think that I will have much more fun coding with kivy than with Qt.
> 

I think the fun factor is an important one to take into account,
especially when contributing to a project in the spare time. However, we
should also throw into the mix the end result for users,
maintainability, documentation and other factors to come up with a good
balance.

If you've got any ideas for the criteria we should evaluate when
choosing the GUI toolkit, please do feel free to add them to the wiki
page, I've added a placeholder text under the comparison table header.

Thanks again for your input!

Cheers,
David.

> Let me know what you think...
> 
> Cheers,
> Stefan
> 
> On Fri, Nov 23, 2012 at 5:01 PM, David Planella
> <david.planella@xxxxxxxxxx <mailto:david.planella@xxxxxxxxxx>> wrote:
> 
>     Al 23/11/12 16:27, En/na Stefan Schwarzburg ha escrit:
>     > Hi David,
>     >
>     > one small question: Your plan was to switch to Qt after this release,
>     > right? Not to develop two GUIs in parallel...
>     >
> 
>     Exactly, that was the idea, a switch and just one single UI.
> 
>     Cheers,
>     David.
> 
> 
>     --
>     Mailing list: https://launchpad.net/~qreator-discuss
>     Post to     : qreator-discuss@xxxxxxxxxxxxxxxxxxx
>     <mailto:qreator-discuss@xxxxxxxxxxxxxxxxxxx>
>     Unsubscribe : https://launchpad.net/~qreator-discuss
>     More help   : https://help.launchpad.net/ListHelp
>

Attachment: signature.asc
Description: OpenPGP digital signature