If you can get a cross platform gui into boost, that meets the standards of boost developers, youll be famous. Many have started and quit after decieding it was to daunting of a task. A modern c++ GUI is probably #1 on the most wanted/needed libraries, IMO,
Tom Brinkman wrote:
If you can get a cross platform gui into boost, that meets the standards of boost developers, youll be famous. Many have started and quit after decieding it was to daunting of a task. A modern c++ GUI is probably #1 on the most wanted/needed libraries, IMO,
I think the basic problem is that if you ask 10 developers what a GUI library should look like, you'll get 11 answers :-P Cheers, -- Joel de Guzman http://www.boost-consulting.com http://spirit.sf.net
Joel de Guzman wrote:
Tom Brinkman wrote:
If you can get a cross platform gui into boost, that meets the standards of boost developers, youll be famous. Many have started and quit after decieding it was to daunting of a task. A modern c++ GUI is probably #1 on the most wanted/needed libraries, IMO,
I think the basic problem is that if you ask 10 developers what a GUI library should look like, you'll get 11 answers :-P
Only 11?
Deane Yang wrote:
Joel de Guzman wrote:
Tom Brinkman wrote:
If you can get a cross platform gui into boost, that meets the standards of boost developers, youll be famous. Many have started and quit after decieding it was to daunting of a task. A modern c++ GUI is probably #1 on the most wanted/needed libraries, IMO,
I think the basic problem is that if you ask 10 developers what a GUI library should look like, you'll get 11 answers :-P
Only 11?
So you should not ask that question. Just make your design decisions, or even better the library. Then you can ask 10 developers if they want it. Only two answers are possible and all the fame may be well deserved ;-) Seriously, has there ever been a real proposal for GUI before? All I have seen are never ending questionnaires on how it should be built. These usually end up in discussions on styles, cascading style sheets, color models, how the point class shall be defined, XML and Unicode support, etc, etc.... No perfect design exist. And there always seems to be something that has to be available before GUI can be started. Someone have to make some calls on this. Bridging existing frameworks seems like one possible direction to go. Maybe if someone came up with a real, far from perfect, but decent proposal, 11 out of 10 would turn out to be fairly forgiving for a not-so-perfect-in-my-eyes design. As pointed out, it is about time something gets moving toward standardization on GUI. Boost.GUI would be a good start. -- Bjørn
On Sunday October 21 2007 13:53, Bjørn Roald wrote:
Maybe if someone came up with a real, far from perfect, but decent proposal, 11 out of 10 would turn out to be fairly forgiving for a not-so-perfect-in-my-eyes design. As pointed out, it is about time something gets moving toward standardization on GUI. Boost.GUI would be a good start.
I think the Java example has shown very nicely that you can't standardize GUIs even if the language is controlled by a single company. And the needs of the C++ community are much more diverse than those of the Java community. Personally, I'm certain that no standardization effort is going to be wasted on something like GUIs for _a very long time_ to come :) Thanks, Marc -- Marc Mutz - marc@kdab.com, mutz@kde.org - Klarälvdalens Datakonsult AB Platform-independent software solutions - www.kdab.com info@kdab.com
Adobe has a "property model" and a "layout model" library (Adam and Eve) which might serve as a starting point and they already have a carbon interface and something for windows (no gtk/qt/wx so far as i know). http://opensource.adobe.com/group__asl__overview.html#asl_overview_adam_and_... Not sure if its useful to build from or contribute to but its out there. Chris On 10/22/07, Marc Mutz <marc@kdab.net> wrote:
On Sunday October 21 2007 13:53, Bjørn Roald wrote:
Maybe if someone came up with a real, far from perfect, but decent proposal, 11 out of 10 would turn out to be fairly forgiving for a not-so-perfect-in-my-eyes design. As pointed out, it is about time something gets moving toward standardization on GUI. Boost.GUI would be a good start.
I think the Java example has shown very nicely that you can't standardize GUIs even if the language is controlled by a single company. And the needs of the C++ community are much more diverse than those of the Java community.
Personally, I'm certain that no standardization effort is going to be wasted on something like GUIs for _a very long time_ to come :)
Thanks, Marc
-- Marc Mutz - marc@kdab.com, mutz@kde.org - Klarälvdalens Datakonsult AB Platform-independent software solutions - www.kdab.com info@kdab.com _______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
You'll always see pushback against GUI libraries. Java's GUI libraries, though flawed, are better than anything in C++. I agree with Bjørn, just make your decisions. No committee. Then see if anyone likes it. It's way to easy to over-design something like this. I imagine the only way it would really work out well is if one lead programmer made all of the hard decisions, and had about 5 different experienced UI or library programmers implementing the innards. Oh, and in your code, you may want to consider allowing the programmer to just define, say, a BOOST_GUI_WIN32 macro, instead of passing around Win32 and GTK objects - as I think you were doing in the samples. Good luck. Jeremy Pack On 10/22/07, Marc Mutz <marc@kdab.net> wrote:
On Sunday October 21 2007 13:53, Bjørn Roald wrote:
Maybe if someone came up with a real, far from perfect, but decent proposal, 11 out of 10 would turn out to be fairly forgiving for a not-so-perfect-in-my-eyes design. As pointed out, it is about time something gets moving toward standardization on GUI. Boost.GUI would be a good start.
I think the Java example has shown very nicely that you can't standardize GUIs even if the language is controlled by a single company. And the needs of the C++ community are much more diverse than those of the Java community.
Personally, I'm certain that no standardization effort is going to be wasted on something like GUIs for _a very long time_ to come :)
Thanks, Marc
-- Marc Mutz - marc@kdab.com, mutz@kde.org - Klarälvdalens Datakonsult AB Platform-independent software solutions - www.kdab.com info@kdab.com _______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman /listinfo.cgi/boost
On 10/22/07, Jeremy Pack <rostovpack@gmail.com> wrote:
You'll always see pushback against GUI libraries.
Java's GUI libraries, though flawed, are better than anything in C++.
I agree with Bjørn, just make your decisions. No committee. Then see if anyone likes it.
I have redesigned lots of things I putted in the email already.
It's way to easy to over-design something like this. I imagine the only way it would really work out well is if one lead programmer made all of the hard decisions, and had about 5 different experienced UI or library programmers implementing the innards.
I'm afraid I may have done that already. But I have some concerns about ODR and using the GUI library by other libraries linked to the same program. And that forced me some design decisions. I also liked the "plug-in" method I used for low-level implementations.
Oh, and in your code, you may want to consider allowing the programmer to just define, say, a BOOST_GUI_WIN32 macro, instead of passing around Win32 and GTK objects - as I think you were doing in the samples.
I didn't document, but there should be a "default" driver. On windows it would be win32, it could also be configurable with preprocessor (though not yet implemented, but *very easy* to do). Passing win32 and gtk objects just override the default.
Good luck.
Thanks!
Jeremy Pack
-- Felipe Magno de Almeida
participants (8)
-
Bjørn Roald -
Chris Fairles -
Deane Yang -
Felipe Magno de Almeida -
Jeremy Pack -
Joel de Guzman -
Marc Mutz -
Tom Brinkman