are on their way to the patches list. Given the mail delay we've been
seeing, they'll take a while to get there. Oh, it turns out there _is_ a
size limit for patches, so it'll need to get approved.
There are still a few warts in the code.
1) One wart is that I needed to make an identifier for the oid for the
"standard" package. The oid in question is 10, and the identifier is
STANDARDPackageId. I think I will change it to StandardPackageId.
The question I have is in which file should I store the define defining
it?
2) Another problem is dealing with the ambiguity between
relation.attribute and package.functionname. The present code does it by
changing scan.l to recognize ${identifier}\.${identifier}, and if the
first identifier isn't a key word, look to see if it is a package (scan
pg_packages for the name). If so, the scanner returns a different token,
PACKID, than IDENT.
I'll see what I can do about moving all of this into the parser, and
defering the pg_packages scan until later.
I think I got rid of all of the debugging comments; please let me know if
I didn't.
Take care,
Bill