Assuming there's no one willing to fix the behavior (and that seems
unlikely given we're in the middle of a 2020-01 commitfest) it makes
sense to at least document the behavior.
That being said, I think the proposed patch may be a tad too brief. I
don't think we need to describe the exact behavior, of course, but maybe
it'd be good to mention that the behavior depends on the type of the
last command etc. Also, I don't think "combined query" is a term used
anywhere in the code/docs - I think the proper term is "multi-query
string".
Any thoughts on Tomas' comments?
Sorry, I did not notice them.
I tried "multi-command", matching some wording use elsewhere in the file.
I'm at lost about how to describe the insane behavior the feature has when
they are used, which is really just an unfixed bug, so I expanded a minima
in the attached v2.
--
Fabien.
diff --git a/doc/src/sgml/ref/psql-ref.sgml b/doc/src/sgml/ref/psql-ref.sgml
index 402c4b1b26..b4a9fc3b60 100644
--- a/doc/src/sgml/ref/psql-ref.sgml
+++ b/doc/src/sgml/ref/psql-ref.sgml
@@ -3781,6 +3781,14 @@ bar
fail after having already displayed some rows.
</para>
+ <note>
+ <para>
+ This feature does not works properly with multi-command
+ (<literal>\;</literal>) queries as it depends on the command
+ types encountered in the string.
+ </para>
+ </note>
+
<tip>
<para>
Although you can use any output format with this feature,
