Removal of Portmanager
rwmaillists at googlemail.com
Sun Jan 13 19:25:47 UTC 2013
On Sun, 13 Jan 2013 17:17:40 +0100
Matthias Andree wrote:
> Am 13.01.2013 17:02, schrieb RW:
> > On Sat, 12 Jan 2013 09:57:16 -0600
> > Bryan Drewery wrote:
> >> On 1/12/2013 6:07 AM, RW wrote:
> >>> "Does not support modern ports features such as MOVED, is
> >>> lacking upstream and active contributions, and does not support
> >>> pkgng. Consider using ports-mgmt/portmaster,
> >>> ports-mgmt/portupgrade or pkgng."
> >>> These seem more like bogus excuses than reasons.
> >>> Portmanager doesn't need MOVED, and the author chose not to
> >>> support it. There's no compelling reason for portmanager users to
> >>> switch to pkgng which may well be the reason no-one has done
> >>> anything.
> >>> The logical time to remove portmanager is when there are no
> >>> supported releases with support for the old package tools - if
> >>> it's not been patched to support pkgng by then.
> >> I do agree that harmless working ports should remain left
> >> untouched. However, portmanager has lacked contributions for years
> >> now,
> > I submitted a bug-fix a few years ago when I found a bug, I haven't
> > submitted any more because I didn't notice any more. Am I to
> > understand that we only permit ports to remain in the tree if they
> > have a minimum level of incorrectness?
> I am very much in support of that view. No half-baked software
> Everyone is free to step up to maintain portmanager - or find (pay)
> someone who does - and bring it up to speed with the recent changes to
> the framework,
Aside from pkgng what changes do do you think it needs?
> rather than endlessly discussing the removal of things
> that got left behind because nobody cared.
How long is it since anyone did any development on awk? Is that going
to go because nobody cares.
> Are you willing to add support for pkg NG to portmanager?
My main objection is that it's been removed so long before it needs to
be. It should have been deprecated six months before it becomes
obsolete, not removed now. Because it rarely gets rebuilt I doubt many
users even knew about this.
> > This is something that people say but never cite any sensible
> > examples. The changes seem to me to be pretty transparent. For me
> > portmanager works better than on the day development ceased. All
> > the problems I've had with updates are traceable to the port system
> > itself.
> How can that be when development has ceased?
Bugs have been fixed since then - development is what causes bugs.
> > To me portmaster and portupgrade's limitations lessen the "overall
> > user experience" more than portmanager's. It's the only one of the
> > three designed to minimise human effort - the other two require
> > much more nursemaiding. We now only have the choice of two tools
> > that place more value on CPU time than my time, and I regard that
> > as a major loss.
> False. For one, portmaster and portupgrade support from-source builds,
> and both deal with setting up proper rebuild order, to reduce your
> personal effort.
Obviously, but that's the bare minimum for an upgrade tool. Portmanager
does more than that.
> > At least you had the luxury of realising it was deprecated. FreeBSD
> > doesn't exactly announce deprecation "on display in the bottom of a
> > locked filing cabinet stuck in a disused lavatory with a sign on the
> > door saying Beware of The Leopard" but it's pretty close.
> > We really need a way of flagging this up for installed packages.
> True enough. Unfortunately, none of the three tools would mention it
> when running it in an "upgrade-all" or "assess-all" mode.
I think it should be included in portaudit.
More information about the freebsd-ports