Re: Tsearch vs Snowball, or what's a source file?

Teodor Sigaev <teodor(at)sigaev(dot)ru> writes:
>> 1) rename FULLTEXT to TEXT SEARCH in SQL command

> Working on it, I found rather obvious undesired side-effect: if TEXT
> becomes a keyword then any output of name of text type becomes
> quoted. Even if TEXT is in unreserved_keyword list.

Yeah, I was aware that would happen. What I've been thinking for some
time is that we ought to fix quote_ident and ruleutils.c to distinguish
"unreserved" keywords from the other ones, and not quote unreserved
ones. The list of unreserved words has gotten longer and more invasive
in every release, and I don't see that stopping soon. It's already
annoying that perfectly ordinary words like "document" and "key" get
quoted.

The main argument I can think of for not doing this is that if a
currently non-reserved keyword becomes reserved in a future release,
then having quoted it would prevent problems with restoring dump
scripts. I don't find this argument very compelling, though; it seems
just as likely that words we don't even have as keywords today will
become reserved later. The only thing I see on the horizon that does
fit that argument is WITH, which we could special-case.

Comments? I'm willing to make this happen if there are no objections.

regards, tom lane