Borland C++-5.5 (free) compiler does not install(?)

[g.patel at wanadoo.fr]

On 29 Apr 2001 00:51:48 +0300, nkarlsso at abo.NO-SPAM-PLEASE.fi (Niclas
Carlsson) wrote:

>I suppose this is to be considered a bug in the linker in question,

Yes.
Thanks for the test app though, it makes things more clear.

> but if
>someone spots an acceptable fix in wine (which Alexandre would accept, that
>is), then do by all means fix it. 

For the algorithm used the 0x80000000 value is a black hole from
which it can get out. I don't see any way for getting Alexandre
Julliard to accept a patch explicitly reserving  this value out of
the range returned by VirtualQuery. I've had a much more generic
patch rejected  as 'too specific' ;-)

The only way I see could be to do a better version of VirtualQuery
that would mark as reserved the memory allocated outside of Wine;
currently you can query the area where glibc is loaded, for example,
and Wine returns it as free; that's not very logical and IMHO a 
good case  could be made that VirtualQuery is bugged from this 
point of view and it should try to detect these memory areas
and report them in VirtualQuery.
That would work only in Linux I guess (I don't know if there is
a portable way to get memory usage ?)

If something like that is done, it would be trivial to reserve the
blackhole area outside of Wine.
Unfortunately I have more important things to do for the next weeks.
If you want to do it, you're welcome :-)

Now if someone sees a simpler way...

Gerard