Re: Fix nptl/tst-cancel7 for non-bash shells

[Jonathan Nieder <jrnieder at gmail dot com>]

Hi,

David Miller wrote:

> For reference, this is the dash fix:
>
> http://git.kernel.org/?p=utils/dash/dash.git;a=commitdiff;h=ee5cbe9fd6bc02f31b4d955606288de36c3d4eab
>
> Then debian, objected to updating to the latest dash code and thus
> getting the fix, because it includes LINENO support and this breaks
> some scripts that use LINENO presence to detect if they can do "bash
> stuff"
>
> So, trying to avoid this bug fix propagating forever, I added a
> configuration option to dash to elide this support:
>
> http://git.kernel.org/?p=utils/dash/dash.git;a=commitdiff;h=4da0e221ba2244921862d46cdeefbc8ab5808eed
>
> But things are still stuck in red tape and still the debian folks,
> for whatever reason, haven't updated their dash package.

I don't know who these "debian" fellows are. ;-)

For what it's worth, in Debian the "dash -c" optimization is
blocked[1] on <http://caml.inria.fr/mantis/view.php?id=5371>.  Debian
7.0 (wheezy) already has dash 0.5.7, which is the latest released
version, but with ee5cbe9f backed out.

I submitted the patch to the dash mailing list that led to commit
ee5cbe9 and would be happy to see this optimization become widespread.
But it is not safe to rely on its presence or its absence.  Whether
the shell can avoid these forks is not a POSIX conformance issue.
POSIX is pretty explicitly agnostic about this.

Hoping that clarifies,
Jonathan

[1] http://bugs.debian.org/436466