Related to the forwarded mail below, the mandoc.db files are also
affected by umask.
Would it make sense for pkg_add and pkg_delete to just force a sane
umask before starting operations?
----- Forwarded message from Marc Espie <es...@nerim.net> -----
From: Marc Espie <es...@nerim.net>
Date: Tue, 10 Mar 2015 13:13:53 +0100
To: Alessandro DE LAURENZIS <just22....@gmail.com>, ports <ports@openbsd.org>
User-Agent: Mutt/1.5.22 (2013-10-16)
Subject: Re: pkg_add and umask [was misc@: Re: Cannot connect to CUPS web
interface in -current]
Mail-Followup-To: Alessandro DE LAURENZIS <just22....@gmail.com>, ports
<ports@openbsd.org>
On Tue, Mar 10, 2015 at 10:54:34AM +0000, Stuart Henderson wrote:
> Moving to ports@ ...
>
> On 2015/03/10 11:32, Alessandro DE LAURENZIS wrote:
> > Hello Stuart,
> >
> > On Tue 10/03/2015 08:28, Stuart Henderson wrote:
> > > Is this while building the port, or just installing from packages?
> >
> > Installing from packages. Isn't that expected?
>
> Not sure.
>
> So:
>
> 1. Database files in /var/db/pkg are affected by umask ("pkg_add moo" with
> umask
> 077, then you can't "pkg_info moo" as a normal user)
>
> 2. Normal installed files from the package are not affected by umask
>
> 3. @sample'd files with an explicit @mode are not affected by umask
>
> 4. @sample'd files *without* an explicit mode (e.g. normal files installed in
> /etc) are affected by umask
>
> ...
>
> 1 could be argued either way, but I think current behaviour is ok.
>
> 2 and 3 seem correct to me
>
> 4 is surprising to me. Marc, is that intentional?
I kindof think 4 should be forbidden or default'd to something sane, indeed.
----- End forwarded message -----
