On Wed, Dec 14, 2011 at 05:01:46PM -0600, Amit Kulkarni wrote:
> On Wed, Dec 14, 2011 at 3:53 PM, Matthias Kilian <[email protected]>
> wrote:
> > - Make it at least startable (it was unusable after the removal of
> > jdk-1.5, and nobody noticed/complained).
>
> now that the ports infra is growing so many ports from 6000 in 4.7 to
> 7200-7500 in 5.1 (estimated), this is going to be a frequent
> occurrence. is there a way we can enforce a basic "make run" to
> atleast checkout if a port runs, somewhat like "make regress". this
> should let us know much faster of potential problems. there won't be
> enough people to test them all anyway. it won't be feasible for all
> ports but for most it would be...
How could this work? People would add some (more or less) random
check-run targets, and with w growing number of ports with such a
target, people would start to ignore failure. We already have a
couple of ports where regress fails, but where the *output* of
regress is still important (hey, lang/ghc, nice to meet you ;-)).
I really doubt that make regress (or something like a check-run
target) would be useful for now.
> > We may as well just remove it from the tree. If there are real
> > users, please speak up!
>
> this is useful for some users, not as if there are 100 other alternatives.
Ok, i'll not delete it ;-)
Ciao,
Kili
