cvs commit: ports/www/firefox Makefile distinfo pkg-descr
pkg-message pkg-plist ports/www/firefox/files firebird.rb
firefox.rb mkdistfile mozconfig.in patch-Double.cpp
eivind at FreeBSD.org
Thu Feb 12 09:56:14 PST 2004
Please take the below as an attempt at being constructive and actually
keeping the discussion on subject. It is somewhat directly formulated
as I don't have time to do much more editing to make it more calm.
On Thu, Feb 12, 2004 at 11:13:25AM -0500, Trevor Johnson wrote:
> > My opinion is that DES is right anyway. This lowers the consistency of
> > the ports tree overall,
> Not really--many other ports do the same thing.
The fact that there are ports other places that have not been updated to
use REINPLACE_CMD is not a reason for introducing a new one. And I
would hope the other ports are moving *away* from using perl also for
replacements, not towards it. The set of people using <= 4.6 was larger
earlier, and they were less out of date earlier, and thus more
reasonable to do hacks for.
[Text deleted in an attempt to keep this as much on subject as
> If someone (you?) have ideas about removing the Perl dependency, that
> would be a really stunning enhancement to the port. As I said it's
> been a huge bugbear.
I don't have enough disk space to even extract it available at the
moment; I may look at it again later (perhaps even tonight.)
However, this is a straw man. The relevant issues (as far as I can
- The replacement of REINPLACE_CMD with PERL -pi is an efficiency hack.
- The efficiency hack increase efficiency on systems of 4.6-vintage and
below. It does this at the expense of consistency with our
conventions, and potentially efficiency on systems newer than 4.6 (due
to larger executable, more shared libs for perl, etc).
Whether increasing efficiency for <=4.6 even has value *when seen
completely apart from the other issues* has been questioned.
- A noticable number of people have said that we value consistency in
the ports tree, and that the change should be reverted. This is
enough to normally trigger a re-evaluation of what is being done.
- There have been no attempt from your side to argue what level of value
is present in the perl changes, beyond saying they remove a
dependency on <= 4.6.
More information about the cvs-all