Avoid using ssize_t type: it's written in POSIX but not in C99 about uthash HOT 4 CLOSED

Hi!
I am not the developer, but I am using utarray in one of my codes. I had one related problem because ssize_t wasn't defined in the includes. As fas as I understand, ssize_t is the signed version of size_t, allowing negative error codes/values.
I would give a try to size_t (in stddef.h), changing the definitions and castings in utarray.h from ssize_t to size_t.
But I am just a user of utarray.h...
Good luck!

from uthash.

I have rechecked the code and the only place where ssize_t is used is in a casting:
Line 217: ... (ssize_t)(a)->icd.sz) ...
and icd.sz is defined as size_t at line 48.
I would comment out the sys/types.h include and would change the ssize_t casting to a size_t one.

from uthash.

@FerCampos
Thanks for your suggestion. This is exactly what I've done. However, from a long-term view, the author may have some ideas why a signed size_t is used instead of a unsigned version. That's why I'm posting here.

from uthash.

I looked at the utarray and utstring code (since ssize_t was also used there) - and it looks like in both cases it really should be size_t. So I changed them to size_t and removed the sys/types.h include in 2eeb5f4. Please reopen if you have any questions. Thanks Hong Xu and Fernando!

from uthash.