FreeTDS Development Group

[Frediano Ziglio <freddy77 AT gmail.com>]

Are you sure is not something related to dependency?

The problem of nmakefile is that is lack lot of stuff. Is not able to
detect additional libraries (openssl, iconv, gnutls and so on). It
does not take version from same sources (configure) so it has to be
manually changed every time. It does not compile all full package
(tests). It does not generate files like versions scripts (for
instance resource versioning for dlls). It does not generate files
from git sources. To add all these feature is quite hard and is mainly
a duplication effort. Actually even cmake+autoconf is a bit a
duplication but first serve very much fine Windows while second is
better for *nixes.

Frediano


2015-06-29 19:01 GMT+01:00 A. Craig West <acraigwest AT gmail.com>:
> I tend to agree, it is pretty much guaranteed to be a compiler bug, but
> fixing compilers is outside the scope of my current project :-)
> I find that the first time I build after a clean, it fails, and if I build
> again it succeeds. I'm not sure if this is actually an error, as it is
> possible there is a broken object file produced from the failed build that
> satisfies the time stamps...
> I will have to look into the cmake option, my nmakefile is working quite
> well, though.
>
> On Mon, Jun 29, 2015 at 2:30 AM, Frediano Ziglio <freddy77 AT gmail.com> wrote:
>
>> 2015-06-29 0:42 GMT+01:00 A. Craig West <acraigwest AT gmail.com>:
>> > I discovered that I require freetds to be buildable as a DLL file under
>> > Windows, so I have modified the Nmakefile to add some new build targets
>> and
>> > directories. After I have tested it a bit, I will port my changes up to
>> > head.
>>
>> We are removing nmake support moving to cmake.
>>
>> > I am also getting a fault in the compiler on one of the files, but have
>> not
>> > investigated the cause yet. It is possible that the issue is already
>> > resolved in head. I am currently building with Visual Studio 2008 command
>> > line.
>>
>> We released quite recently possibly problem is still there. Or could
>> be you called a function with wrong arguments. Technically "a fault in
>> the compiler" looks like a compiler bug to me.
>>
>> > Jas anybody else dealt with similar issues lately? It seems to have been
>> a
>> > while since anybody tried to build DLL's, although it turned out to be
>> > pretty easy to implement...
>> > -Craig
>>