Re: [Breakage] Git v2.21.0-rc0 - t5318 (NonStop)

From: Jeff King <peff@peff.net>
Date: Feb 9, 2019, 23:29 UTC
Message-ID: <20190209232928.GA7135@sigill.intra.peff.net>
In-Reply-To: <38f594ed-0787-aeb5-d34f-2f4775243eb7@kdbg.org>

On Sat, Feb 09, 2019 at 09:39:43AM +0100, Johannes Sixt wrote:

Show 6 quoted lines

> > Great. Since it sounds like you're preparing some patches to deal with
> > /dev/zero elsewhere, do you want to wrap it up in a patch as part of
> > that?
> 
> Please do not use yes to generate an infinite amount of bytes. Our
> implementation of yes() in test-lib.sh generates only 99 lines.

Ah, thanks. That doesn't matter here, but it would for the other patches under discussion.

Show 6 quoted lines

> Perhaps do this.
> [...]
>  	dd of="$objdir/info/commit-graph" bs=1 seek="$zero_pos" count=0 &&
> -	dd if=/dev/zero of="$objdir/info/commit-graph" bs=1 seek="$zero_pos" count=$(($orig_size - $zero_pos)) &&
> +	printf "%0*d" $(($orig_size - $zero_pos)) 0 | tr 0 '\0' |
> +		dd of="$objdir/info/commit-graph" bs=1 seek="$zero_pos" &&

Using stdin instead of the tmpfile is nice, and shouldn't have any problems. I do think your printf suggestion looks nice, but I wondered if it might run into portability issues (not because of anything in particular, but I often find that the more clever a shell solution, the more likely we run into obscure problems).

But if it works everywhere, that's fine by me.

-Peff