On Fri, Sep 26, 2008 at 11:25 AM, Brendan Jurd <direvus@gmail.com> wrote:
> One way to tidy this up would be to re-implement the meridiem markers
> using the seq_search functions, i.e., make it work like the day and
> month names. This would make it easy to accept any flavour of marker,
> and the error messages thrown for bogus input would then be the same
> as those for bogus day and month names.
Yeah if we seq_search then it should be a pretty easy conversion. so +1
However that still leaves the original complaint around (at least IMHO):
select to_timestamp('AN', 'AM');
ERROR: invalid AM/PM string
select to_timestamp('11:47 PM 27 Sep a2008', 'HH:MI PM DD Mon YYYY');
ERROR: invalid value for "YYYY" in source string
Now arguably most to_timestamp calls are going to be short so i can
easily look for the YYYY in my string and see what I did wrong (and
DETAIL: provides Value must be an integer in the second case)... So
really maybe thats good enough I dunno... Personally I find the output
of my dumb patch to be better than both above:
select to_timestamp('AN', 'AM');
ERROR: invalid AM/PM string for 'AN'
And we could improve that by only showing node->key->len chars. And
make that work for both AM/PM and the others YYYY etc
Then again maybe its not worth it?
