[PATCH 0/3] git-clone fails when current user is not in /etc/passwd

From: Jeff King <peff@peff.net>
Date: Dec 10, 2015, 21:32 UTC
Message-ID: <20151210213228.GB29055@sigill.intra.peff.net>
In-Reply-To: <20151210204034.GA29055@sigill.intra.peff.net>

On Thu, Dec 10, 2015 at 03:40:35PM -0500, Jeff King wrote:

Show 9 quoted lines

> > All true.  The adding of "(none)" in add_domainname() I used as an
> > excuse to make the function stop barfing is a symptom coming from
> > the above.  That one should die when asked to do a strict thing
> > (there is a corresponding kludge to do strstr "(none)" to cover it
> > up, which is even uglier X-<).
> 
> Yes, I have always hated that. It seems like we should just be able to
> carry a "strict" flag (both from xgetpwuid() and from the "(none)" hack)
> along with ident_default_*, and barf at the right time if it is not set.

I don't think we want to pass down a "be strict" flag to the low-level code filling in default_{name,email}. We might be strict in one call, and non-strict in another within the same program. Worse, we actually call ident_default_name() even when we don't actually want the name (we could fix that, too, but it just adds more cases to the code).

So here's my solution, which instead carries the "is it bogus" flag along with the default strings.

  [1/3]: ident: make xgetpwuid_self() a static local helper
  [2/3]: ident: keep a flag for bogus default_email
  [3/3]: ident: loosen getpwuid error in non-strict mode

-Peff