Thread: Re: [BUGS] BUG #3767: tsearch2 index creation fatal crash

Re: [BUGS] BUG #3767: tsearch2 index creation fatal crash

From
Tom Lane
Date:
"Thomas H." <me@alternize.com> writes:
> i saw my mistake. i misinterpreted the examples to show the possibility
> to convert *any* text by using casting to tsvector as an alternative to
> using to_tsvector :)

Yeah, the examples in section 12.1.2 are not actually good practice:
both sides of the match should usually get normalized before comparison,
and these examples don't do that.

I'm not sure that putting in to_tsvector calls there would be an
improvement, though, since at that point we haven't introduced
to_tsvector.

Thoughts anyone?

            regards, tom lane

Re: [BUGS] BUG #3767: tsearch2 index creation fatal crash

From
Bruce Momjian
Date:
Tom Lane wrote:
> "Thomas H." <me@alternize.com> writes:
> > i saw my mistake. i misinterpreted the examples to show the possibility
> > to convert *any* text by using casting to tsvector as an alternative to
> > using to_tsvector :)
>
> Yeah, the examples in section 12.1.2 are not actually good practice:
> both sides of the match should usually get normalized before comparison,
> and these examples don't do that.
>
> I'm not sure that putting in to_tsvector calls there would be an
> improvement, though, since at that point we haven't introduced
> to_tsvector.
>
> Thoughts anyone?

Yep, I saw that chicken and egg problem in the docs when I was reviewing
it long ago.  I never came up with a solution either.  We would have to
introduce configurations a lot earlier, but how would you explain them
when you don't know what a tsvector is.

--
  Bruce Momjian  <bruce@momjian.us>        http://momjian.us
  EnterpriseDB                             http://postgres.enterprisedb.com

  + If your life is a hard drive, Christ can be your backup. +

Re: [BUGS] BUG #3767: tsearch2 index creation fatal crash

From
Tom Lane
Date:
Bruce Momjian <bruce@momjian.us> writes:
> Tom Lane wrote:
>> I'm not sure that putting in to_tsvector calls there would be an
>> improvement, though, since at that point we haven't introduced
>> to_tsvector.
>>
>> Thoughts anyone?

> Yep, I saw that chicken and egg problem in the docs when I was reviewing
> it long ago.  I never came up with a solution either.

OTOH, I see that the very next sentence mentions to_tsquery without
defining it.  So maybe it wouldn't be materially less readable if we
just changed the examples to use to_tsquery and to_tsvector instead
of literal-casting.  Either way, there is something going on that
doesn't meet the eye, and we might as well write an example that is
actually OK practice rather than bad practice.

            regards, tom lane

Re: [BUGS] BUG #3767: tsearch2 index creation fatal crash

From
Tom Lane
Date:
>> I'm not sure that putting in to_tsvector calls there would be an
>> improvement, though, since at that point we haven't introduced
>> to_tsvector.

Uh ... wait a moment ... stop the presses.  Thomas is looking at
beta1 documentation.  CVS HEAD handles this a bit better I think:
http://developer.postgresql.org/pgdocs/postgres/textsearch-intro.html#TEXTSEARCH-MATCHING

So the first thing is to go bug the www team about updating the online
8.3beta docs, which I shall do forthwith.

            regards, tom lane