> So I would base this discussion on the premise "bytea stores binary data"
> (insert examples).
>
> Some stylistic issues:
>
> bytea => <type>bytea</type>
>
> NULLs => zero bytes/bytes of value zero ("NULL" is too overloaded)
>
> 'non-printable' => <quote>nonprintable</quote>
>
> MUST => <emphasis>must</emphasis>
>

Here's a patch against *CVS tip* to address Peter's comments. Please let
me know what you think!

Joe Conway

needs it.

Seems it is getting compileable...

Marko Kreen

and error handling behavior.

Bradley McLean

libpq.sgml

I was trying (but gave up) to cross-reference back to the input escape
table in the User's Guide, but could not get the documentation to
compile with a cross-book xref (missing IDREF error). Can a cross-book
xref be done?

 Joe Conway

should be all.

Marko Kreen

failures on FreeBSD.  This patch replaces uint -> unsigned.

This was reported by Daniel Holtzman against 0.4pre3 standalone
package, but it needs fixing in contrib/pgcrypto too.

Marko Kreen

Operators", plagiarized shamelessly from the "String Functions and
Operators" section. There were enough differences that it made sense (at
least to me) to give this its own section instead of cramming it in with
normal string functions. This way I could also make the examples
relevant, which is particularly important for bytea.

One thing I think worth mentioning: while documenting the trim()
function I realized that I never implemented the bytea equivalent of
rtrim and ltrim. Therefore, the 'leading' and 'trailing' forms of trim,
available with text, are not available with bytea (I'd be happy to
correct this, but since it would require an initdb, I guess not until
7.3) -- the submitted doc accurately reflects this.

I will look for other areas of the docs that need mention of bytea, but
any guidance would be much appreciated.

--

Here's a second bytea documentation patch. This one significantly
expands the "Binary Data" section added by Bruce recently.


Joe Conway

update configure.in for b3 ...

earlier, make error checks more uniform.

display (with a typemod) and function arg/result type display (without
a typemod).

and REVOKE FROM PUBLIC: the latter is not the same as 'revoke from all
users', but the ref page blurred the difference.

with other data types, per disucssion.  Encoding issue still open.

    o they sometimes returns a result garbage string appended.
    o they do not work if client encoding is different from server
      encoding

Overall a really nice job here, Peter ...

the interactive docs.

enforce a limit on who can connect to databases other than their own.
From a recent discussion in pg-admin.