Portmaster ignoring +IGNOREME??

[Mark Ovens]

Doug, it's happened again:

[...]

===>>> Updating package dependency entry for each dependent port

===>>> Re-installation of mjpegtools-1.9.0.r3 succeeded

===>>> Returning to check of ports depending on icu-3.8.1
===>>> Launching child to update mozilla-1.7.13_3,2
         icu-3.8.1 >> mozilla-1.7.13_3,2

===>>> Port directory: /usr/ports/www/mozilla
         ===>>> This port is marked DEPRECATED
         ===>>> The mozilla team has shifted the development branch of 
mozilla to  seamonkey, this port is outdated and has many 
vulnerabilities.  Please consider using www/seamonkey or www/firefox 
instead.

         ===>>> If you are sure you can build it, remove the
                DEPRECATED line in the Makefile and try again.

===>>> Update for mozilla-1.7.13_3,2 failed
===>>> Aborting update

===>>> pkg-message for gnokii-0.6.22,1
================================================================================

[....]

It honoured the +IGNOREME during the recursive ``make config'' but 
failed during the build.

In this case, unlike the last, it was using the same version of 
portmaster (2.1) throughout. The command I ran was:

    portmaster -BudRr icu-3.8.1/

The only common thing between the two instances of this happening is 
that a large number of ports were being rebuilt/upgraded. Is this 
significant?

As this run took the best part of 24 hours it is a bit of a pain. The 
reason mozilla has a +IGNOREME is because it is only required by 
openoffice, which also has a +IGNOREME. The only workround I can think 
is to move the mozilla and openoffice directories out of /var/db/pkg 
while I run portmaster, then back again afterwards.

This also highlights a potential problem with portmaster. because it 
prints out all the pkg-message files at the end, and the summary list, 
there is the potential for the reason for the abort to have scrolled out 
of the shell's scrollback buffer (it wasn't far off in this instance) 
thereby making it impossible to find out which port failed - in fact the 
only reason I knew that something had failed was because I knew that 
some Xorg stuff had to be rebuilt and it didn't appear in the summary.

Can I suggest the following enhancements?

1. That portmaster reports the port that failed (and why, if possible) 
in the summary.

2. That the pkg-message file output it written to a file rather than the 
screen (with a message where the file is) to avoid the potential problem 
of any failures - and indeed some of the pkg-message output - being lost 
from the scrollback buffer. Maybe this could be an option rather than 
the standard behaviour?

Regards,

Mark