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

[Kees Dekker]

>Thanks for writing up that patch. I installed the attached patches, which I hope 
>addresses a few of the issues you raised. I'll comment below on the issues that 
>the attached patches don't address.

I will check the patches at end of this week, or next week. So please give me some time.
I will also need some more time to read all feedback. Thanks for providing this.

On beforehand: telling that a compiler is wrong is probably true/valid from developers or clean-coding perspective, 
but it is a very cumbersome job to change compiler behavior. I understand your feedback, but I can't do much with it.

In most cases, I'm using the pragmatic approach, and understand that I've to live with the (probably) odd behavior 
of some compilers. It makes life too complicated to argue to all vendors that they must change their compiler. 
E.g. HP does not spend much time anymore to their compiler (because HPUX is almost dead?) and arguing with Microsoft 
about standards is certainly something that will never succeed.  We have no other choice than continue using 
the existent compilers (and in most cases, Visual Studio is not that bad). Even if I can convince vendors to 
adjust their compiler, it takes months or years before these changes become available (I'm ignoring for case of 
simplicity the fact that standards are not always 100% clear, and leave some room for vendor's own thought about 
an implementation).

Regards,
Kees