Christoph M. Becker
Christoph M. Becker
> Perhaps there are more issues that need to be properly fixed. Apparently not. :)
I think that this patch would change the behavior: `s` is unsigned, and the `-` is relevant, and the code is just relying on underflow, what is even specified behavior...
> error C2039: 'const_pointer': is not a member of 'std::allocator' Try with an older C++ standard; that failed for me with C++20, but worked with the default (for me C++14).
Note that building against early pre-releases is certainly a good idea to catch issues (either necessary changes to extension code, or, like in this case, upstream issues), but please do...
It seems pdo_sqlsrv 5.12.0 builds fine against PHP 8.4.0alpha2 with the same modifications.
In a perfect world, these warnings should be resolved, but as far as I can tell, these are level 2 and 3 warnings; if you want them suppressed, remove the...
Note that I've pushed a commit regarding the proper calling convention (otherwise won't build on Windows, and might have issues on other platforms). And I have **force-pushed** to get the...
Oh, and thank you for the PR! This might be getting somewhere. :)
This still needs some work: - [x] failing tests should let CI fail (~~unrelated to this PR, but needs to be fixed~~ that was actually related to this PR; forgot...
While the clean-up turned out rather different to what I had in mind (I completely changed the return mechanism approach, back to swapping the exit handler), this looks good to...