I've had a buildfarm module testing cross-version pg_upgrade for a while now. It's fairly stable - it considered there's a success when the diffs from the original and the converted dumps match the following perl table:

                                 #target    source
        my $expected_difflines = {
                                  HEAD => {
                                           REL9_0_STABLE => 1715,
                                           REL9_1_STABLE => 643,
                                           REL9_2_STABLE => 741,
                                          },
                                  REL9_2_STABLE => {
                                                    REL9_1_STABLE => 0,
                                                    REL9_0_STABLE => 1085,
                                                   },
                                  REL9_1_STABLE => {
                                                    REL9_0_STABLE => 1085,
                                                   }
                                 };


A compressed copy of the actual diffs it finds is attached for those who are interested.

However, code changes can obviously change the numbers, so I don't want to hardcode them in a Module.

I was thinking that a file in the pg_upgrade directory containing theses numbers would be a sensible3 way to go. The file would just contain lines like (for master):

   REL9_0_STABLE 1715
   REL9_1_STABLE 643
   REL9_2_STABLE 741


When we branch the REL9_3_STABLE branch we'll add a line like this to the master branch file:

   REL9_3_STABLE 0


The buildfarm module could parse this file and test accordingly. If we change things then all we'd need to do to get the buildfarm green again would be to adjust this file accordingly.

Thoughts? Better suggestions?

cheers

andrew

Attachment: dumpdiff.txt.gz
Description: GNU Zip compressed data

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

Reply via email to