Peter Eisentraut <peter.eisentraut@enterprisedb.com> writes:
> On 2020-11-24 16:46, Tom Lane wrote:
>> Maybe we should decide that completeness is more important than keeping
>> the line to some arbitrary width. But it's operating as designed.
> I think we should get rid of the truncating. Otherwise, there is no way
> to actually get the full information, is there? (Other than pg_dump or
> manual catalog queries.)
That'd be okay with me. It's always seemed a little odd that we do that
for attrdefs but not anything else.
A bit of checking with git blame says that the habit came in with
the very first version of describe.c, in
commit a45195a191eec367a4f305bb71ab541d17a3b9f9
Author: Bruce Momjian <bruce@momjian.us>
Date: Thu Nov 4 21:56:02 1999 +0000
Major psql overhaul by Peter Eisentraut.
which has
+ /* Info */
+ cells[i*cols + 2] = xmalloc(128 + 128); /* I'm cutting off the default string at 128 */
+ cells[i*cols + 2][0] = '\0';
+ if (strcmp(PQgetvalue(res, i, 4), "t") == 0)
+ strcat(cells[i*cols + 2], "not null");
+ if (strcmp(PQgetvalue(res, i, 5), "t") == 0) {
+ /* handle "default" here */
+ strcpy(descbuf, "SELECT substring(d.adsrc for 128) FROM pg_attrdef d, pg_class c\n"
+ "WHERE c.relname = '");
So that looks very much like the truncation was an expedient thing
to do to work with a fixed-size result buffer, rather than something
that was chosen to improve the user experience.
regards, tom lane