cvs commit: ports/lang/tclX Makefile

[Andrew Pantyukhin]

On 5/3/07, Kris Kennaway <kris at obsecurity.org> wrote:
> On Thu, May 03, 2007 at 01:40:52AM +0400, Andrew Pantyukhin wrote:
> > On 5/3/07, Kris Kennaway <kris at obsecurity.org> wrote:
> > >On Wed, May 02, 2007 at 07:57:58PM +0000, Mikhail Teterin wrote:
> > >> mi          2007-05-02 19:57:58 UTC
> > >>
> > >>   FreeBSD ports repository
> > >>
> > >>   Modified files:
> > >>     lang/tclX            Makefile
> > >>   Log:
> > >>   Work-around the deficiencies in FreeBSD package-building cluster,
> > >>   which causes the regression-tests to fail by disabling the regression
> > >>   tests.
> > >>
> > >>   Unbreak -- the build was never "broken" for the usual (non-jail)
> > >>   systems.
> > >
> > >Thanks.  Just for the record, building in a jail is a supported
> > >configuration for all users, so it's something that is required to
> > >work for all ports.
> >
> > So can we please actually write it down somewhere?
>
> I am not sure of an appropriate place to record this fact.
> Suggestions?

Anywhere in the Porter's Handbook, naturally. I'll be
updating docs this summer, you can probably leave it to
me if you don't find this urgent. This thread will
remain starred in my inbox until resolved.

> > I've never
> > heard that we support that, but it's great if we do. The only
> > thing a ports developer needs is a comprehensible way to set
> > up a jail for testing purposes (comprehensible = without
> > trying to parse tinderbox/pointyhat source files himself).
>
> It's trivial, and described in the jail manpage.  Basically you just
> installworld into some directory, mount devfs and ports, then run
> jail(8) and build things "as normal".

Yes, I always have a few jails around built the "canonical"
way described in the manpage, but I always have trouble
building some ports. Sometimes it's just perl and hundreds
of other ports install OK, sometimes dozens of them report
problems.

It clearly requires investigation from my side to tell you
more, but it seems it's not that simple to recreate
pointyhat's success and failure cases. Maybe there are some
subtle details about pointyhat's jails that need to be added
to jail(8) manpage.

Thanks!