This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: bootstrap problems
On Thu, Jun 07, 2001 at 04:49:40PM +0200, Gabriel Dos Reis wrote:
> Marc Espie <espie@quatramaran.ens.fr> writes:
> | Code that's added to the compiler that does more sanity checks of internal
> | rules is good.
> But if such code stops bootstrapping, there is no benefit since no
> work can be persued. Such sanity checks are definitely good if they
> come along with the cure symptoms they reveal.
You can't expect people to be able to fix every platform, especially for
somewhat obscure stuff like mips. Does Jakub even have access to a mips,
for starters ?
It looks like he ran into the same problem a few months back.
Did a mips guy jump in and fix the problem so that he could commit his
patch ? nope. so, the `good patch' got lost.
Everything still bootstraps.
And there's still a time-bomb ticking in the mips code...
I think there's a middle ground.
Guy proposes an extra check. Check doesn't pass bootstrap on a platform
because the platform is buggy.
So you publicize the patch, and people from that platform have some
deadline to find out what's going with their platform (say, one month).
If, after that deadline's past, no-one's bothered to move to fix the
problem, patch is committed. Bootstrap breaks. And platform people can
still fix the problem.
Sound unreasonable ?