Listing the ports that will be upgraded in advance

Clemens Lang cal at macports.org
Thu Feb 19 06:52:36 PST 2015


Hi,

----- On 19 Feb, 2015, at 15:07, René JV Bertin rjvbertin at gmail.com wrote:

> Possible, but from what I could tell the only that was broken was the registry
> entry. Doxygen contains only a binary, basically,  and any issues due to -n
> should haven been caught by the rev-upgrade step before I started on the
> dependent port.
> Right?

No. I can easily come up with a case that's not covered:

Consider a dep chain A -> B -> C. B links statically against C for some reason.
A uses a feature from B which in turn uses a feature from C. Now, we change the
configuration of C (let's say we enable reading and writing of file format X).
Normally, we'd revbump B to recompile it against the new C to pick up the
feature so it's available in A. If you use -n, you'll miss that, rev-upgrade
won't notify you and your system will behave unexpectedly. In the end, you might
notice your A doesn't read and write X files, and you file a bug with us, which
we end up tracking down to the problem after what would probably be a lot of
wasted effort.

tl;dr: do NOT use -n. It is considered harmful, and it is considered harmful for
a reason.

-- 
Clemens Lang


More information about the macports-dev mailing list