Re: [wxhaskell-devel] CPP to FFI

2013-06-14 Thread Blair Archibald 
Ah I see. I'm still looking through the wxHaskell code to see if there is
any work I think I can do (as a beginner!) and thought id give this a
mention.

I agree that SWIG is a great tool! Hopefully they get their summer of code
working.

Many thanks,
Blair


On 14 June 2013 15:02, Jeremy O'Donoghue  wrote:

> From the website...
>
> "To use fficxx, you write a Haskell model of the C++ public interfaces
> and fficxx generates both a C wrapper and associated haskell functions and
> type classes which reflect specified model of the C++ interfaces. It is
> currently the user’s responsibility to specify a correct model of the C++
> interfaces, because fficxx does not presently check for model correctness."
>
> This is the opposite of what you need for wxHaskell - the ideal scenario
> is to parse the C++ headers and use them to generate Haskell bindings
> automatically. Writing correct Haskell representations of the complete
> wxWidgets API will be very painful and is likely to be error-prone (it is
> quite a large API).
>
> There are a few key things that any automated binding generator needs to
> handle:
> * Callback functions - these are used extensively in wxWidgets;
> * Memory management - the approach in wxHaskell today is not completely
> satisfactory and there are quite a few memory leaks.
>
> I believe that the only game in town which is really sufficiently mature
> is SWIG - wxPython bindings are already generated by SWIG, so it is known
> to be fit for purpose. If the Haskell Qt binding generator goes in this
> direction, it may well be what is needed.
>
> Regards
> Jeremy
>
>
> On 14 June 2013 10:27, Eric Kow  wrote:
>
>> Seems like something the author would be very interested in testing on.
>> I think the wxHaskell paper talks about how it's done here
>> http://research.microsoft.com/pubs/66810/wxhaskell.pdf
>>
>> On 13 June 2013 20:43, Blair Archibald 
>> wrote:
>> > Hi guys,
>> >
>> > Did anyone see this new C++ foreign function generator mentioned in the
>> > haskell weekly news.
>> >
>> > http://ianwookim.org/fficxx/
>> >
>> > At the moment the lack of documentation puts me off, but something like
>> this
>> > could be the end of some of our generation problems. In my mind it's
>> great
>> > that people are working on this sorta thing!
>> >
>> > I might look into this deeper at the weekend and see if I could use it
>> to
>> > generate wx bindings.
>> >
>> > Many thanks,
>> > Blair
>> >
>> >
>> >
>> >
>> >
>> --
>> > This SF.net email is sponsored by Windows:
>> >
>> > Build for Windows Store.
>> >
>> > http://p.sf.net/sfu/windows-dev2dev
>> > ___
>> > wxhaskell-devel mailing list
>> > wxhaskell-devel@lists.sourceforge.net
>> > https://lists.sourceforge.net/lists/listinfo/wxhaskell-devel
>> >
>>
>>
>>
>> --
>> Eric Kow 
>>
>>
>> --
>> This SF.net email is sponsored by Windows:
>>
>> Build for Windows Store.
>>
>> http://p.sf.net/sfu/windows-dev2dev
>> ___
>> wxhaskell-devel mailing list
>> wxhaskell-devel@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/wxhaskell-devel
>>
>
>
--
This SF.net email is sponsored by Windows:

Build for Windows Store.

http://p.sf.net/sfu/windows-dev2dev___
wxhaskell-devel mailing list
wxhaskell-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/wxhaskell-devel

Re: [wxhaskell-devel] CPP to FFI

2013-06-14 Thread Jeremy O'Donoghue 
>From the website...

"To use fficxx, you write a Haskell model of the C++ public interfaces and
fficxx generates both a C wrapper and associated haskell functions and type
classes which reflect specified model of the C++ interfaces. It is
currently the user’s responsibility to specify a correct model of the C++
interfaces, because fficxx does not presently check for model correctness."

This is the opposite of what you need for wxHaskell - the ideal scenario is
to parse the C++ headers and use them to generate Haskell bindings
automatically. Writing correct Haskell representations of the complete
wxWidgets API will be very painful and is likely to be error-prone (it is
quite a large API).

There are a few key things that any automated binding generator needs to
handle:
* Callback functions - these are used extensively in wxWidgets;
* Memory management - the approach in wxHaskell today is not completely
satisfactory and there are quite a few memory leaks.

I believe that the only game in town which is really sufficiently mature is
SWIG - wxPython bindings are already generated by SWIG, so it is known to
be fit for purpose. If the Haskell Qt binding generator goes in this
direction, it may well be what is needed.

Regards
Jeremy


On 14 June 2013 10:27, Eric Kow  wrote:

> Seems like something the author would be very interested in testing on.
> I think the wxHaskell paper talks about how it's done here
> http://research.microsoft.com/pubs/66810/wxhaskell.pdf
>
> On 13 June 2013 20:43, Blair Archibald  wrote:
> > Hi guys,
> >
> > Did anyone see this new C++ foreign function generator mentioned in the
> > haskell weekly news.
> >
> > http://ianwookim.org/fficxx/
> >
> > At the moment the lack of documentation puts me off, but something like
> this
> > could be the end of some of our generation problems. In my mind it's
> great
> > that people are working on this sorta thing!
> >
> > I might look into this deeper at the weekend and see if I could use it to
> > generate wx bindings.
> >
> > Many thanks,
> > Blair
> >
> >
> >
> >
> >
> --
> > This SF.net email is sponsored by Windows:
> >
> > Build for Windows Store.
> >
> > http://p.sf.net/sfu/windows-dev2dev
> > ___
> > wxhaskell-devel mailing list
> > wxhaskell-devel@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/wxhaskell-devel
> >
>
>
>
> --
> Eric Kow 
>
>
> --
> This SF.net email is sponsored by Windows:
>
> Build for Windows Store.
>
> http://p.sf.net/sfu/windows-dev2dev
> ___
> wxhaskell-devel mailing list
> wxhaskell-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/wxhaskell-devel
>
--
This SF.net email is sponsored by Windows:

Build for Windows Store.

http://p.sf.net/sfu/windows-dev2dev___
wxhaskell-devel mailing list
wxhaskell-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/wxhaskell-devel

Re: [wxhaskell-devel] CPP to FFI

2013-06-14 Thread Eric Kow 
Seems like something the author would be very interested in testing on.
I think the wxHaskell paper talks about how it's done here
http://research.microsoft.com/pubs/66810/wxhaskell.pdf

On 13 June 2013 20:43, Blair Archibald  wrote:
> Hi guys,
>
> Did anyone see this new C++ foreign function generator mentioned in the
> haskell weekly news.
>
> http://ianwookim.org/fficxx/
>
> At the moment the lack of documentation puts me off, but something like this
> could be the end of some of our generation problems. In my mind it's great
> that people are working on this sorta thing!
>
> I might look into this deeper at the weekend and see if I could use it to
> generate wx bindings.
>
> Many thanks,
> Blair
>
>
>
>
> --
> This SF.net email is sponsored by Windows:
>
> Build for Windows Store.
>
> http://p.sf.net/sfu/windows-dev2dev
> ___
> wxhaskell-devel mailing list
> wxhaskell-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/wxhaskell-devel
>



-- 
Eric Kow 

--
This SF.net email is sponsored by Windows:

Build for Windows Store.

http://p.sf.net/sfu/windows-dev2dev
___
wxhaskell-devel mailing list
wxhaskell-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/wxhaskell-devel