On Thu, Jul 18, Yang Tse wrote:

> On Thu, Jul 18, Daniel Stenberg wrote:
>
>> But since I reverted that due to a problem that was reported with some other
>> windows compiler, is there a chance that this or pieces of this is (also)
>> compiler dependent and not OS dependent?
>
> Yes it is possible. I'll investigate very soonish and see what comes out.

Yes, finally there was an issue with the WIN32 Memory Tracking code,
triggered with borland compiler. Wide strdup code support was
protected with WIN32 preprocessor definition, but this wasm't enough
given that borland does not define wcsdup() prototype when compiling a
non-unicode target.

Commit 6bcacff1 fixes the issue, ensuring that WIN32 MemoryTracking
wide strdup code support is only 'visible' when both WIN32 and UNICODE
are defined.

Additionally there were some issues in Borland makefiles, now
addressed with commit 2e5b3168. Verified that it builds clean with
Borland's free edition 5.5.1.

With these two commits in place, next daily snapshot should build
clean OOTB at least for 5.5.1. I don't have other versions in order to
test personally.

HTH

-- 
-=[Yang]=-
-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette:  http://curl.haxx.se/mail/etiquette.html