ports/69024: Please portlint [databases/mysqltcl]

[Mark Linimon]

On Wed, 14 Jul 2004, Wesley Shields wrote:

>  On Tue, Jul 13, 2004 at 10:26:00PM -0400, Adam Weinberger wrote:
>  
>  > Are you testing these changes or just submitting them?
>  
>  Not all of the 20 I just submitted have been tested (mainly because
>  it would take forever to build quite a few of these)

I'm sorry, but from where I sit, untested changes in PRs are worse
than useless -- they can lead lead submitters into committing untested
changes into the ports collection, which, since Murphy's Law does hold,
*will* get them hate-mail immediately.  The committers, themselves,
will have to test these changes one-by-one, so someone will have to
do the work.  It would be best if there were no surprises during that
step.

portlint is a good tool, but it is not a perfect one.  It uses
heuristics to try to 'guess' what usages that might be bugs in port
Makefiles.  It is right more often than not, but blindly follwing
its advice will quickly lead to broken ports.  This isn't just my
opinion or me trying to be negative, this is me speaking from
experience, and I have the arrows in the back to prove it.

Please, please, please, if you want to help us make the ports collection
the best it can possibly be, set up a ports jail on a 5.x box and start
working through the build errors (see pointyhat.freebsd.org or the
summaries on portsmon.firepipe.net) -- that is where our most pressing
needs exist.  If you have an amd64 machine or a sparc or alpha, so much
the better.

This will help you to get started on learning where the significant
errors lie, and to understand more about whether to take the output of
portlint as correct or incorrect.  Thanks.

mcl