Re: PATCH: Configurable file mode mask

David Steele <david@pgmasters.net>

From: David Steele <david@pgmasters.net>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Stephen Frost <sfrost@snowman.net>, "Tsunakawa, Takayuki" <tsunakawa.takay@jp.fujitsu.com>, Pg Hackers <pgsql-hackers@postgresql.org>, Adam Brightwell <adam.brightwell@crunchydata.com>
Date: 2017-03-13T17:32:33Z
Lists: pgsql-hackers
Hi Tom,

On 3/13/17 1:13 PM, Tom Lane wrote:
> ... oh, and now that I've actually looked at the patch, I think it's
> a seriously bad idea to proceed by removing the mode parameter to
> PathNameOpenFile et al.  That's basically doubling down on an assumption
> that there are NO places in the backend, and never will be any, in which
> we want to create files with nondefault permissions.  That assumption
> seems broken on its face.  It also makes the patch exceedingly invasive
> for extensions.

I think it's a bad idea to have the same parameters copied over and over
throughout the code with slight variations (e.g. 0600 vs S_IRUSR |
S_IWUSR) but the same intent.

In all cases there is another version of the function (added by this
patch) that accepts a mode parameter.  In practice this was only needed
in one place, be_lo_export().  I think this makes a pretty good argument
for standardization/simplification in other areas.

Thanks,
-- 
-David
david@pgmasters.net


Commits

  1. Allow group access on PGDATA

  2. Refactor dir/file permissions

  3. Revert "Add basic TAP test setup for pg_upgrade"

  4. Add basic TAP test setup for pg_upgrade