[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: balsa build help



Glenn Maynard <g_deb@zewt.org> writes:

> On Thu, Jan 24, 2002 at 11:00:19AM +0000, Jules Bean wrote:
> > AIUI, the only reason that this is considered bad practice is the
> > instability (and non-uniformity) of the C++ ABI.
> 
> I don't think that effects the portion of the API exported to C.
> 
Yes, however, how do you handle the SONAME?

> > > I would suggest reporting a bug against libpspell, demanding a split
> > > in, say libpspell and libpspell++, or similiar.
> 
> That's completely unnecessary.  (And, er, "demanding" it would be a
> quick way to find myself ignored ...)  It would certainly do nothing to
> solve this problem.
> 
Please read my previous post. I thought pspell was part C/part C++, so
it would be proper to split it. I didn't know it was written in C++.

But, anyhow, a split would still avoid the SONAME problem mentioned
above. And s/demand/suggest/g.

> > How could the split sensibly be performed, given that pspell is
> > written in C++?
> 
> Often, libraries are written in C++, then a C wrapper is written around
> them to give a C interface; but that wrapper itself has to be in C++, to
> be able to use the C++ library.  There's nothing wrong with that.
> 
Yeah, but it's cleaner IMO to have the C wrapper in its own library,
linked against the C++ lib. That way you can manage the SONAMES
independently, which is probably what you want, since the C++ ABI is
much more likely to change than the C ABI.

Andy
-- 
Andreas Rottmann         | Dru@ICQ        | 118634484@ICQ | a.rottmann@gmx.at
Georg-Rendlweg 28        | A-5111 Bürmoos | Austria       | Europe
http://www.8ung.at/rotty | GnuPG Key: http://www.8ung.at/rotty/gpg.asc
Fingerprint              | DFB4 4EB4 78A4 5EEE 6219  F228 F92F CFC5 01FD 5B62



Reply to: