Hi Bruce,
I think there has been some confusion about the bit/bit varying
stuff I did. I initially
sent in a bit type that was non-SQL compliant, with the suggestion that
this may be
useful to stick into contrib.
Then several people remarked that it would be better to have an SQL
compliant type, so
I wrote the code to do all the manipulations, roughly based on
src/backend/utils/adt/varchar.c
This can evidently not be used as a standalone routine, but needs to be
properly
integrated into the database as a type. Unfortunately I don't really
know an awful
lot about the internals of postgres, and got a bit stuck trying to
figure out how to
integrate a new type correctly.
Now I see that this second set of routines has ended up in the contrib
directory,
where they are quite useless. In my opinion it would be best to have no
bit-type
in the contrib directory and have the proper SQL compliant one properly
integrated
into postgres, but I'm going to need some hints to do that correctly.
So my question is how we should proceed on this. Am I right in assuming
that there
should be no major objections to adding an SQL type to postgres, or
should I go
through another iteration on the mailing list?
Thanks for any suggestions,
Adriaan
