please revert graphics/xfig r354029

Steve Kargl sgk at troutmask.apl.washington.edu
Sat May 31 18:29:16 UTC 2014


On Sat, May 31, 2014 at 12:31:28PM -0500, Mark Linimon wrote:
> On Sat, May 31, 2014 at 08:09:36AM -0700, Steve Kargl wrote:
> > I forgot I had the DOCS option unset as it was unset ages ago
> > and updates have always worked.  The question is "why are changes
> > to a port committed without proper testing?"  Yes, "proper
> > testing" should include testing of the effects of (un)setting
> > individual Makefile options.
> 
> The number of combinations is huge.

There are 3 options for graphics/xfig.  It takes all of 5 minutes
to build and install xfig on a 5 year old laptop.  The options
are mutually exclusive, so one needs to build and install the
port 4 times (i.e., a whole 20 to 25 minutes).

> 
> It's just not feasible.
>

I'm not advocating that portmgr should set up the testing
on the FreeBSD cluster.  The testing should be done by the 
individual maintainers.  If s/he includes 3 (or 2 dozen)
options in the Makefile, then s/he should test those options
when s/he changes/updates the port.  If it is too inconvenient
or too labor intensive to test the options (i.e., 2 dozen options),
then perhaps the options aren't too terribly important and should
be removed.
 
> Even if it were, we have 2042 ports PRs (171 or so are about staging),
> and those ought to be our priorites IMHO.

Yes, those should be the priorities.  Hopefully, any change
to a port (that addresses one of these PRs) is properly tested.

-- 
Steve


More information about the freebsd-ports mailing list