Daniel Stenberg wrote:
> On Sat, 16 May 2009, Yang Tse wrote:
>
>> The drawback of this solution is ABI breakage for Win32 platforms.
>
> I'm not really the man to tell about the exact solution, but this memory
> leak is indeed worthy to fix and I do not think we should just wait for
> or depend on a future MS fix.
>
> If this mentioned approach is deemed the best way, and in my view it
> seems to be, then an ABI breakage on win32 is worth it I think.
>
>> If this is the way to go, should this be included in current cvs
>> c-ares version or once next release takes place?
>
> If this is the way, then we should fix it asap in CVS so that people can
> start testing it out and tweaking apps to work with it. And then
> consider a release with this change included to happen soon.
>
Hello,
So, it's:
- stop LoadLibrary("iphlpapi.dll") and thus breaking Win32
cross-compilation setup (whatever that is, I've never used it)
versus
- breaking ABI compatibility and force all users add two function calls
in their applications
I cannot understand why you prefer to break ABI compatibility.
-- Best regards, Vlad DinulescuReceived on 2009-05-18