On 7/2/20 9:44 AM, Jeremy Schneider wrote:
On 7/2/20 09:28, David G. Johnston wrote:
The status quo prevailed since no-one chose to contribute further
arguments for change and the original patch was retracted. What kind
of "bike-shedding" (which seems to be used incorrectly here) would you
expect?
The bike-shedding comment reflects my sense of humor, which is evidently
so dry that it can be entirely un-detectable. :) But overall I do
think there's room for input from more people. Might not change the
outcome, I'm just curious if there are more people who'd have thoughts
to offer.
I'm not sure hint text would be the right course, as the hint wouldn't
make sense in the context of other clients and I don't know if we have a
mechanism now for the server to change it's error response based on
which client is being used.
One other thought occurred to me after sending this email - does
ON_ERROR_ROLLBACK=interactive differentiate between syntax errors and
other errors? Thinking about how users approach SQL, this feels
significant. I'd happily want to let users at my company retry after
syntax errors, but I'd want them to inspect any other error more closely.
An error is an error:
\set ON_ERROR_ROLLBACK interactive
test(5432)=# begin ;
BEGIN
test(5432)=# slect 1;
ERROR: syntax error at or near "slect"
LINE 1: slect 1;
^
test(5432)=# select 1;
?column?
----------
1
(1 row)
test(5432)=# select 1/0;
ERROR: division by zero
test(5432)=# select 1;
?column?
----------
1
(1 row)
-Jeremy
--
Jeremy Schneider
Database Engineer
Amazon Web Services
--
Adrian Klaver
adrian.klaver@xxxxxxxxxxx
