Re: [PATCH v3 3/4] parse_commit(): handle broken whitespace-only timestamp

From: Jeff King <peff@peff.net>
Date: Apr 27, 2023, 16:46 UTC
Message-ID: <20230427164611.GA1976451@coredump.intra.peff.net>
In-Reply-To: <ad4f927f-607a-5391-ce50-a342d09934d6@gmail.com>

On Thu, Apr 27, 2023 at 12:55:26PM +0100, Phillip Wood wrote:

Show 15 quoted lines

> On 27/04/2023 11:11, Phillip Wood wrote:
> 
> > > +    test_commit 1234567890 &&
> > > +    git cat-file commit HEAD >commit.orig &&
> > > +    sed "s/>.*/>    /" <commit.orig >commit.munge &&
> > > +    ws_commit=$(git hash-object --literally -w -t commit
> > > commit.munge) &&
> > > +    sed "s/>.*/>   $(printf "\013")/" <commit.orig >commit.munge &&
> > 
> > Does the shell eat the '\v' when it trims trailing whitespace from the
> > command substitution (I can't remember the rules off the top of my
> > head)?
> 
> Having looked it up, the shell trims newlines but not other whitespace so
> this should be fine.

Yep. I also wondered if some sed versions might complain. But it's really not that much more exotic than a tab, so it's probably OK.

-Peff