Re: [PATCH] Proposal for HIDDEN/INVISIBLE column

Le 18/10/2021 à 22:36, Alvaro Herrera a écrit :
> On 2021-Oct-18, Gilles Darold wrote:
>
>> Le 18/10/2021 à 18:54, Alvaro Herrera a écrit :
>>> I suggest to look for output test files that are being massively
>>> modified by this patch. I think those are likely unintended:
>>>
>>>> diff --git a/src/test/regress/expected/collate.icu.utf8.out b/src/test/regress/expected/collate.icu.utf8.out
>>>> diff --git a/src/test/regress/expected/collate.linux.utf8.out b/src/test/regress/expected/collate.linux.utf8.out
>>>> diff --git a/src/test/regress/expected/compression.out b/src/test/regress/expected/compression.out
>>>> diff --git a/src/test/regress/expected/xml.out b/src/test/regress/expected/xml.out
>>>> diff --git a/src/test/regress/expected/xmlmap.out b/src/test/regress/expected/xmlmap.out
>> My bad, thanks for the report Alvaro. New patch version v3 should fix that.
> Hmm, the attachment was 500kB before, about 30% of that was the
> collate.*.out files, and it is 2.2 MB now. Something is still not
> right.

Right I don't know what I have done yesterday, look like I have included
tests output autogenerated files. However I've attached a new version v4
of the patch that include the right list of files changed and some fixes:

- Allow a table to have all columns unexpanded, doc updated.

- Add a note to documentation about use of ROWTYPE when there is an
unexpanded column.

- Fix documentation about some sgml tag broken.

About ROWTYPE generating an error when SELECT * INTO or RETURNING * INTO
is used with unexpanded column, I have kept things like that because it
is the normal behavior. I have checked on others database engine and
this is the same.

--
Gilles Darold