This is the continuation of a discussion that started 9 months ago, and
went like this: I proposed that, rather than living in a chaos of Perl
date/time modules, we should have, as a goal, to get to one commonly
accepted foundation module on top of which all other date functions
would be built. Rich Bowen kindly answered that he _had_ contacted the
date/time modules authors before, and that it was "abundantly clear that
there is no desire to cooperate on a unified set of Date/Time modules"
(both messages should be in the archives).

I found Rich Bowen's answer convincing, but I also happened to discover
another possible way out of the chaos. However, during these last nine
months, my interests have shifted, and I have no more intention to work
with Perl dates. But I thought that, for completeness, I should tell you
about the possible alternative in case anyone finds any value in it.
(However, the information I'll give might be somewhat outdated, since I
don't know whether there's been any relative change in 5.8 or so.)

If one of the date/time modules were standard, and all the other ones
were optional, this would be a strong indication that the standard
module is the preferred one. Alternatively, the Perl FAQ could encourage
using a specific module rather than others. If this could be done, then
everyone could continue to use their own choice of chaotic modules,
whether for backwards compatibility or for love of chaos, but Perl
newcomers and Perl programmers writing new programs would probably get
used to the standard module. That's the basic idea.

Now I don't know how top management works (i.e. how it is decided which
modules are standard or what goes into the FAQ), but it is interesting
that Larry Wall once showed some interest on the date/time problem. He
seems to believe that the standard date/time Perl interface is not good,
and he proposed a new interface. You can find his message at
http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2000-01/msg00241.html.
Matt Sergeant subsequently implemented Larry Wall's idea into the
Time::Piece module, which you can install and man for more information.

I think that Time::Piece's interface is simple enough and powerful
enough, and that it is good foundation for building more complex date
handling (such as date parsing) on top of it. And due to the community's
and Larry Wall's interest in the Perl date problem, it could become
standard.

Now I have a strange feeling that, for some of you, this is hardly any
news; and if it is not, then there is a problem with the mailing lists,
that is, the problem is being discussed on some list which I had failed
to find, despite the fact that I had searched much 9 months ago, only to
find [EMAIL PROTECTED], which still leaves me the feeling that someone's
discussing these things elsewhere.

Reply via email to