On Wed, Jan 1, 2014 at 5:08 PM, Gary V. Vaughan <g...@gnu.org> wrote:
...
> The change (back ticks instead of $(...) in the no-submodule-changes rule)
> looks like a harmless way to fix the bug to me.
>
> Okay to push?
Hi Gary,
I have been unable to reproduce that failure so far.
E.g., this via grep (which I have just updated to use the latest from gnulib):
make no-submodule-changes SHELL=/bin/bash V=1
works fine for me using this version of bash and others:
GNU bash, version 4.2.45(2)-release (i386-apple-darwin12.3.0)
Many people run rules that depend on the one you're proposing
to change, so I suspect that something unusual about your environment is
at the root of this. Besides, you know how I feel about backticks vs $()
syntax (we should almost always prefer the latter), so am very reluctant
to make such changes without understanding the root cause.
