Time library discussion (reprise)

[Peter Simons]

Simon Marlow writes:

 > Presumably, without TAI support, the library will be
 > unable to do accurate calendar calculations at the second
 > level [...]

As I understand it, calendar time calculations into the
future are _always_ inaccurate, because UTC is not
continuous. Either you'll get an incorrect "duration" or
you'll get an incorrect "point in time". It can't be helped.

The C++ time library of the Boost effort has a very
interesting discussion of this topic, available here:

    http://boost.org/libs/date_time/doc/Tradeoffs.html

Pardon me, if this has been brought up already.

As for implementing a time library in Haskell, we might be
able to base the effort on one of the existing API designs
-- or even one of the existing implementations? There are,
among others:

 - The Boost.Time library mentioned above.

 - A proposed API extension for the ISO C 200X standard:
   http://www.cl.cam.ac.uk/~mgk25/time/c/

 - Proposed ISO C 200X Calendar Library Functions:
   http://david.tribble.com/text/c0xcalendar.html

 - Dan Berstein's libtai: http://cr.yp.to/libtai.html

Surely there is _something_ out there that can be re-used?

Peter