To: Henrik Storner <storner AT olicom DOT dk>
Cc: djgpp AT sun DOT soe DOT clarkson DOT edu, eliz AT is DOT elta DOT co DOT il
Subject: Re: New time routines 
Date: Wed, 31 Aug 94 09:14:34 +0200
From: eliz AT is DOT elta DOT co DOT il

>             The zoneinfo files just _have_ to be present, and in the
> right place, too! Is it reasonable to expect that "ordinary users (tm)" 
> have such a setup ?

DJGPP is not the first package to depend on proper setup for proper
operation.  What's so hard about doing pkunzip -d -o, even for
"ordinary users (tm)"?  And if it's distribution of production
software you are talking about, you can always tell your clients
to put a full POSIX TZ string.  Or just have the ``posixrules''
file alone installed, and EST5EDT will be interpreted like you
want.  Voila! no need for 300K of files!

> My intention with the routine I posted was not to replace tzset(), but
> rather to provide a means for being able to handle incomplete TZ specifica-
> tions, even when the zoneinfo files are missing.

The problem with your routine is that it arbitrarily adds DST rules
to the TZ setting.  These might be *mostly* right, but they cannot be
*always* right, because there are all kinds of strange countries out
there (like my little corner, for example) with most weird DST
legislature which can easily defy any ``simpe'' solution (just
look at the contents of israel file which DJ distributes, and
you'll understand what I mean).  That's *exactly* the cause that
whoever wrote tzset() didn't do what you suggest in the first place.
But if you want a simple solution like you posted, you only need a
single file called ``posixrules'' which does what you suggested if
TZ is incomplete.  A single file is, I think, not too much of a burden
to install.

	Eli Zaretskii