Broken due to CVS branching? .bki has wrong info for build

On Jul 17, 2010, at 12:44 PM, Selena Deckelmann <selenamarie@gmail.com> wrote:

Hi!

I pulled the latest from the git repo, I got this error on initdb:

creating template1 database in testdb/base/1 ... initdb: input file
"/usr/local/pg90/share/postgresql/postgres.bki" does not belong to
PostgreSQL 9.0devel

The problem was having '9.1' instead of '9.0' in the first line of the .bki.

Are you sure you have a clean checkout? git clean -dfx, maybe?

...Robert

Selena Deckelmann <selenamarie@gmail.com> writes:

I pulled the latest from the git repo, I got this error on initdb:

creating template1 database in testdb/base/1 ... initdb: input file
"/usr/local/pg90/share/postgresql/postgres.bki" does not belong to
PostgreSQL 9.0devel

The problem was having '9.1' instead of '9.0' in the first line of the .bki.

Yeah, I ran into that too. The makefiles don't have a dependency that
forces postgres.bki to be rebuilt when you update the major version in
configure.in. Just delete src/backend/catalog/postgres.bki and
rebuild/reinstall. (Actually, as of CVS HEAD it shouldn't be a problem
anymore because of yesterday's changes in pg_proc.h.)

I'm not sure whether it's worth adding an explicit dependency to cover
this case. It only comes into play at major version boundaries.

regards, tom lane

On Sat, Jul 17, 2010 at 10:51 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:

Selena Deckelmann <selenamarie@gmail.com> writes:

I pulled the latest from the git repo, I got this error on initdb:

creating template1 database in testdb/base/1 ... initdb: input file
"/usr/local/pg90/share/postgresql/postgres.bki" does not belong to
PostgreSQL 9.0devel

The problem was having '9.1' instead of '9.0' in the first line of the .bki.

Yeah, I ran into that too.  The makefiles don't have a dependency that
forces postgres.bki to be rebuilt when you update the major version in
configure.in.  Just delete src/backend/catalog/postgres.bki and
rebuild/reinstall.  (Actually, as of CVS HEAD it shouldn't be a problem
anymore because of yesterday's changes in pg_proc.h.)

I'm not sure whether it's worth adding an explicit dependency to cover
this case.  It only comes into play at major version boundaries.

Ok, that makes sense.

I rebuilt just now, and it worked fine.

Thanks!
-selena

--
http://chesnok.com/daily - me