home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!mcsun!uknet!uos-ee!mcs.surrey.ac.uk!(null)
- From: css1he@mcs.surrey.ac.uk (Hani El-Sakkout)
- Newsgroups: comp.os.ms-windows.programmer.tools
- Subject: Re: BC++ 3.1 Memory Alloc Problem
- Message-ID: <1992Jul21.103609.21192@EE.Surrey.Ac.UK>
- Date: 21 Jul 92 10:36:09 GMT
- References: <1992Jul20.094558.20975@EE.Surrey.Ac.UK> <1992Jul20.220044.14222@thunder.mcrcim.mcgill.edu>
- Sender: news@EE.Surrey.Ac.UK
- Organization: University of Surrey, Guildford, England
- Lines: 26
-
-
- Thanks for your follow-up, Leying..
-
- |> You use 'new' to allocate memory. The one you use is borland classlib's
- |> class Object new. It redefined. Actually, it is the same with system
- |> 'new'. Windows convert it to LocatAlloc().
-
- I *think* this is true only when using the small and medium memory
- models, but in the Large memory model (the Borland's manuals say that)
- heap allocations are from the Global heap. The Borland's technical
- support people also confirmed this by saying that my module definition
- file heap-size settings should have absolutely no effect because there
- is no local heap in the Large memory model!
-
- Which brings me to another question - do the module definition
- settings used during the production of library and DLL files
- have any effect on the final .EXE? I doubt it, but you never
- can tell..
-
- Anywaym, an update on my original query..
-
- Unless I have overlooked something my present struggles lead me
- to suspect that there is a SERIOUS FLAW with the DLL OWL/Container
- class library memory allocation system under Borland C++ 3.1.
-
- I hope I'm wrong..
-