2007/1/24, Marius Vollmer <[EMAIL PROTECTED]>:
"ext Kalle Vahlman" <[EMAIL PROTECTED]> writes:
> 2007/1/24, Marius Vollmer <[EMAIL PROTECTED]>:
>> Yep, the C++ bindings are unfortunately pretty low priority for me.
>> As far as language bindings go, C++ doesn't need them (people can just
>> use the C API from C++).
>
> You can write a C module and import that to Python too, no need for bindings.
> Oh, wait, that WAS the binding. IMO the situation is very much the
> same for C++, though it's of course much easier to include C in C++
> than in Python.
It's not entirely the same. Python and other higher level languages
have different type systems and manage memory differently than C.
Exactly.
Cleanly wrapping a C API for such a language needs support from the C
API most of the time. Gtk+ has tons of this support, and hildon-fm
should have it, too.
C++ on the other hand can use C APIs unwrapped.
...but will lose any ease of development or automatic memory
management that make up the whole point of writing C++. So you can
just as easily say that no need for bindings, since you can write in C
anyway.
I am happy to heavy significant Perl trickery into the build system to
generate run-time type information about enumerations. But changing
typedef struct { ... } foo;
to
typedef struct _foo foo; struct foo { ... };
for the benefit of C++ bindings? Come on. That seems ill-motivated
to me.
Granted, this particular thing is something that the wrappers should
handle, but this wasn't the only one.
> It still doesn't invalidate the need for bindings.
There is no need for C++ bindings. It is nice to have them, tho.
That's like saying "I don't care about the bindings".
I can understand the lack of time to make things work for C++, but I
don't think ignoring trivial(ish) _patches_ just because you can use
some other language is nice.
Making the application developer wrap lists and such to STL containers
for example is not loving the developer. And anyone developing /
maintaining a library _should_ love the application developer,
otherwise the API will serve the library implementation not the user.
> I guess we'll thank you for it. Eventually.
:-) Do it now! I applied the patch.
Thanks!
--
Kalle Vahlman, [EMAIL PROTECTED]
Powered by http://movial.fi
Interesting stuff at http://syslog.movial.fi
_______________________________________________
maemo-developers mailing list
maemo-developers@maemo.org
https://maemo.org/mailman/listinfo/maemo-developers
