[tz] proposed changes for Win32 and a improved mktime() algorithm

Kees Dekker Kees.Dekker at infor.com
Mon May 29 11:52:34 UTC 2017

>Yes, and tzalloc() and one or two which are just _z ... if it has a 'z'
>suffix in there somewhere, then it is relevant.  The 'z' implies that
>it has a timezone_t as a parameter (which tzalloc() makes from a TZ string).
>Oh, there's also tzfree() of course - the inverse of tzalloc().



>There should not be a lot, and I cannot think of anything of a different
>nature than you have seen already.

>One issue may be the 'r' suffix part (thread safe) - but I assume windows
>has some way to make threaded programs, and hence to make thread safe library

Windows at default (although with some exceptions, especially to POSIX like functions) is
multi-threaded aware. But as our application is not, it is not a problem. I will check
how I can use these _z and/or _rz functions.

I will first try to make an isolated patch proposal for time_succesive(), as asked by Paul (for mktime() performance).

More information about the tz mailing list