RE: [ft-devel] FreeType compatibility with C++ managed extensions

[Graham Asher]

David,

I think it's best to stick with my original decision to withdraw the
proposal. I don't really like these things myself, but at first (and I see I
was wrong) it looked like a big win for a small effort.

If you're talking about my compromise idea to allow the public FreeType
headers to be #included by managed C++ code (in the same way that they can
be successfully #included by ordinary C++ code now) then, yes, it would be
nice if we could change 'generic' and 'internal', but I understand the
difficulty of changing the API.

<<<<<<
There might a solution by using some ugly processors tricks, with a small
price to pay for managed-C++ users which should use one or two distinct
identifiers (e.g. "generik" instead of "generic")
>>>>>>

Personally I hate this sort of respelling, but it's just an aesthetic
problem. Donald Knuth seems to like it - do you remember his "noad" objects
which were a special type of node?

<<<<<<
Do you have a complete list of problematic keywords ?
>>>>>>

Not yet - I gave up investigating this after a little while. For the
compromise solution only identifiers in the headers are a problem. I *hope*
it's only generic and internal, but I'm not sure.

Oh, there's another thing which if it lasts, and is not removed as a
transitional feature, will provide a work-around: you can compile the C++
code that #includes FreeType with /clr:oldsyntax (I think that's it) and it
uses old keywords with leading __ (and I don't think generic exists at all
in the old syntax).

Best regards,

Graham