Please do not reply directly to this email. All additional comments should be made in the comments box of this bug report.
Summary: Review Request: sepostgresql - Security-Enhanced PostgreSQL
https://bugzilla.redhat.com/show_bug.cgi?id=249522
------- Additional Comments From kaigai@kaigai.gr.jp 2007-08-27 18:43 EST ------- (In reply to comment #48)
For example, the PL and -test subpackages Require: postgresql-server, which would mean you'd have to duplicate those as se-specific versions.
Indeed, you are correct. It's a demerit of using Conflict: tag.
I don't see the reason why you can't make it install parallel files with different names (sepostgres etc). The argument that some error messages include the program name seems a bit silly, and it's been awhile since there were any hard dependencies on the executable name.
I don't say it is not possible, but I was anxious about this change may invoke unexpected problems.
pg_ctl and initdb invoke the hardcoded "postgres", so we have to patch them in the sepostgresql package at least.
Just I confirmed the /usr/bin/postgres source. Several codes seems invoke the hardcoded "postgres", but most of them are enclosed by "#ifdef EXEC_BACKEND", so it seems to me the binary name does not give us any significant affect.
OK, I try to rename the binary, and confirm its works.
Rather than Conflicts: postgresql-server, I wonder whether you shouldn't be trying to Require: postgresql-server = %{version} so that you can share whichever files are in common, instead of shipping duplicates.
Maybe, we can omit any libraries (like utf8_and_euc_jp.so) and timezone data.
Thanks,