On Wed, Jun 13, 2007 at 02:08:47PM -0400, Blair Sadewitz wrote:
> How do you define "does not work"?  If you mean "the build fails",
> then yes, that does hold true for a lot of the larger packages.

Build fails is one class. Build being incomplete or incorrect is
another. Typical examples of rules that break apart as soon as parallel
builds are attempted are:
- using *.o without full dependencies, esp intra-directory depends
- yacc/lex rules

The point is just because it build doesn't mean it won't fail in obscure
ways. IMO we had enough such reports already that the current approach
is simply inacceptable.

Joerg