Re: [HACKERS] ALTER TABLE ... ALTER COLUMN ... SET DISTINCT

[Dimitri Fontaine]

Hi,

Finally some update on the patch.

Le 18 juil. 09 à 20:55, Robert Haas a écrit :

This is one of the things that I hate about the requirement to post
context diffs: filterdiff, at least for me, strips out the git tags
that indicate the base rev of the patch.

Yes, and as I didn't have the time to install filterdiff I've attached a revision of your patch in git format which adresses the problem I mentioned, in a tarball also containing raw notes, tests, and regression.{out,diffs}. mqke check is failing on opr_sanity test in what looks like an ordering issue, but I didn't feel confident enough to adapt the .out to force the regression into passing.

 - style issue, convert at PQgetvalue() time

Ah, good catch.

Fixed in the updated version, which uses a float4 variable in pg_dump and strtod() rather than atof. This last point is maybe overkill as I'm using: tbinfo->attdistinct[j] = strtod(PQgetvalue(res, j, i_attdistinct), (char **)NULL);

What about adding the following before the switch, to do like surrounding

code?
       Assert(IsA(newValue, Integer) || IsA(newValue, Float));

Not a good plan.  In my experience, gcc doesn't like switch ()
statements over enums that don't list all the values, unless they have
a default clause; it masterminds by giving you a warning that you've
"inadvertently" left out some values.

                default:
                        elog(ERROR, "unrecognized node type: %d",
                                 (int) nodeTag(newValue));

I've left this part alone but still wonders about this, which is a new user visible error message kind:

Given your revised version I'll try to play around with ndistinct behavior

and exercise dump and restore, etc, but for now I'll pause my work.

dim=# alter table foo alter column x set distinct 1;
ALTER TABLE
...
dim=# alter table foo alter column x set distinct 0;
ALTER TABLE
dim=# analyze verbose foo;
INFO:  analyzing "public.foo"

ANALYZE

 attname | attdistinct
---------+-------------
 x       |           0
(1 row)



Regards,
--
dim

I failed to have 0 to allow for analyze to compute a value again, as shown in the raw notes in attachement: INFO: "foo": scanned 4 of 4 pages, containing 1000 live rows and 0 dead rows; 1000 rows in sample, 1000 estimated total rows dim=# select attname, attdistinct from pg_attribute where attrelid = 'foo'::regclass and attname = 'x'; What I understand from the doc part of your work contradicts what I see here... Will mark as Waiting on Author, as I need Robert to tell me what to do next.

ndistinct.tgz
Description: Binary data

-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers