Hi Viktor
> Did you try the other method?
Which other method. May be I missed something, please
repost it.
<<<
Surely they are both much more elegant than xhb cstructs,
as this latter passes binary structures around in _strings_,
so all it takes is a silly .prg level mistake to corrupt C level
data.
>>>
I agree. But in the absense of another way this is the
only workaround available at the moment.
<<<
Yes, such MT problems are notoriously difficult to find,
so just because you didn't see them doesn't mean
they aren't there, all it takes is some MT app which does
some things actually in parallel + bad luck and the app
crashes / corrupts memory or data. The structure code
simply isn't obeying some basic MT programming rules,
see my NOTE in the ChangeLog:
---
; TOFIX: - hbwhat/wincore.prg
- xhb/cstruct.prg
Needs reviewing and either mutex protection
for STATICs or conversion to THREAD STATIC.
---
>>>
100% agreed.
<<<
IMO a better solution would be to recode these in a non-generic
way (each CALLBACK gets a separate wrapper), instead of
using this generic callback solution which isn't portable.
>>>
I am looking into that aspect, though not difficult, may
be not as efficient as this one is. Windows only sends
<hWnd> to the wndproc and I need to tie WvgWindow()
class object with WvgWindow():hWnd. I have an idea. I have
to make experiments first...
Regards
Pritpal Bedi
--
View this message in context:
http://www.nabble.com/SF.net-SVN%3A-harbour-project%3A-10378--trunk-harbour-tp22133738p22149487.html
Sent from the Harbour - Dev mailing list archive at Nabble.com.
_______________________________________________
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
