Document Upper Limit for NAMEDATELEN in pgsql 9.5+
Postgresql 9.5+ may now fail to compile if NAMEDATELEN has been set
absurdly large (In my case, 384).
The file src/backend/utils/adt/levenshtein.c does a static assert on
"NAMEDATALEN <= MAX_LEVENSHTEIN_STRLEN" with MAX_LEVENSHTEIN_STRLEN
currently set to 255.
When using a gcc version that does not support static asserts, the error
message looks like:
- "levenshtein.c:104: error: negative width in bit-field
'static_assert_failure'"
It would be great to be mention this new upper limit on NAMEDATALEN in the
documentation.
Here are some places I've looked at:
- src/include/pg_config_manual.h
- http://www.postgresql.org/docs/9.5/static/sql-syntax-lexical.html
- http://www.postgresql.org/docs/9.5/static/runtime-config-preset.html
- https://wiki.postgresql.org/wiki/Todo
While on the subject of NAMEDATALEN, there is a second undocumented place
where NAMEDATALEN is defined:
- src/interfaces/ecpg/include/sqlda-native.h
--
Kevin Day
Kevin Day <thekevinday@gmail.com> writes:
Postgresql 9.5+ may now fail to compile if NAMEDATELEN has been set
absurdly large (In my case, 384).
The file src/backend/utils/adt/levenshtein.c does a static assert on
"NAMEDATALEN <= MAX_LEVENSHTEIN_STRLEN" with MAX_LEVENSHTEIN_STRLEN
currently set to 255.
Hmm. I'm not sure whether 384 is "absurdly large", but I do wonder
why the levenshtein code gets to dictate limits on NAMEDATALEN at all.
Or to put it even more bluntly, I'm not sure that there is anything
whatever about MAX_LEVENSHTEIN_STRLEN that is well thought out. Why not
rip it out and put some CHECK_FOR_INTERRUPTS tests into the loops instead?
regards, tom lane
--
Sent via pgsql-docs mailing list (pgsql-docs@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-docs