On Fri, Oct 07, 2016 at 07:53:13PM +0000, Guenter Milde wrote:

> It does not suffice to test whether compilation runs without an error.
> This does in no way guarantee that the output is OK.

I see. I suppose it comes down to whether we think that the proposed
changes have a better chance of fixing something than breaking
something. As Jean-Pierre pointed out, we already have a case where such
changes fixed something. I don't know enough to estimate what the chance
is that the proposed changes lead to an error that is not reported by
compilation and that can only be seen by manual inspection.

Scott

Attachment: signature.asc
Description: PGP signature

Reply via email to