Re: [Devel] 2.1.0 branch created

[Antoine Leca]

Happy New Year to all of you here.

Hi Boris,

Boris Letocha wrote:
> 
> > >So the question really boils down to:
> > >
> > >  A - is there enough interest in a C++ version of the library
> > >      (and more especially a C++ native API)
> > >
> > >  B - would doing so drastically reduce our development time
> > >      (man-years ??)
> > >
> 
> 1.) So your first solution C++ without STL doen't give you too much
> things better except "exceptions" which helps you to make library easily
> any leak safe ... ( unfortunately exceptions in MSVC++ are painfuly slow )

I tend to agree here.
I believe C++ certainly does have advantages, but they need to engage
yourself in things that are not (yet?) mature enough to be consistently
implementated, at least for a system library.
Put in another way, using templates or STL or RTTI right now has a high
probability that the result may be catastrophic in term of performances
on "some" exotic platforms, which will quickly drive Freetype as a
g++-only library, which in turn is not IMHO a good idea.

 
> 2.) Next is usage of MLib. I think (I am not sure) that usage of longjmp
> is not as portable as C code without longjmp.

I do not think so. longjmp did exist since the advent of C in the 70's,
and even if it was not consistently implemented in the 80's, it is part
of the ANSI standard and now (remember, ANSI standard is about 15 years
old now), almost every C implementation does have a reasonable longjmp.


> Also IMHO if you use longjmp in C++ you will not get C++ stack cleaning.

This should be tested. I have no information, but if you feel this is
a problem (where?), it should be investigated.


> 3.) And leave it in "plain" C - for us it is no problem, but for you
> I think it is pain and really lost of time.

What do you mean?


Antoine