winapi_check question...

[Greg Turner]

On Thursday 31 October 2002 12:54 pm, Patrik Stridvall wrote:
> Greg wrote:
> > On Thursday 31 October 2002 09:20 am, Patrik Stridvall wrote:
> > > Ignore it.
> >
> > OK.  You promise it won't screw up the actual documentation bot?
>
> Promise? Hmm. All code have bugs you know and winapi_check is
> designed to be ad hoc and thus inherently buggy... :-)
>
> But seriously, I wouldn't worry about it.

BUT I CANT SLEEP!!!!  (j/k, it's already forgotten)

> Anyway. A real fix to the code you be nice.

but I /am/ being nice ;)

> Presumable the type CLIENT_CALL_RETURN haven't
> been declared properly.

no can do, it's that way because of an obscure gcc vs. msvc call 
convention mismatch for function calls with struct return-types.  

anyhow, CLIENT_CALL_RETURN is just a stupid union for casting between 
longs and ptr's, so, to me, it's just not worth the effort to 
understand the problem.  Basically, I just did what Alexandre told me 
to, and called it "fixed" :)

A real fix probably involves some crazy wrapper/thunking technique, 
assembly code, wierd gcc command-line options, an extention to the 
.spec format, and/or obscure #pragma's -- quite possibly beyond my 
abilities, and definitely beyond my level of motivation, given the 
(presumed) non-severity of this particular instance of the problem... 

maybe, however, there are some more pressing reasons to solve this, in 
which case I'd be happy to look at it a bit more carefully (but still 
skeptical of my ability to solve it).

-- 
gmt

"The purpose of government is to rein in the rights of the people"
 --President Bill Clinton, MTV interview, 1993