cvs commit: ports/cad/admesh Makefile

[Matthias Andree]

Am 13.08.2011 15:37, schrieb Dmitry Marakasov:
> * Doug Barton (dougb at FreeBSD.org) wrote:
> 
>> We've had this discussion about 6 times now.
> 
> Maybe we should stop doing things that raise such discussions then?

No.  Such poking the sleeping is necessary to get action taken at all.

Evidently it hadn't happened until the time the respective ports in that
run had been deprecated.

>> The community consensus is that we need to cull dead ports from
>> the tree in order to reduce the maintenance burden and allow for
>> more flexibility.
> 
> "Dead" means it doesn't build or doesn't work. Which exactly of
> these "unfetchable" ports doesn't build or doesn't work?

It's irrelevant.  Unmaintained ports of any kind are a danger to the end
users.  They might miss crucial (security or critical) patches, and you
can't judge the quality of software by "it builds and we mirrored an
obsolete version years ago".

Typically this situation arises if there are no users of a port left,
else the complaints come much earlier.  And if there are no users,
culling a port no matter whether your local culture considers it "dead"
is not a loss except that we might some day fall below 20,000 ports.
Which is neither likely nor undesireable. :)

> That is strange definition of "dead". Does it stop being dead if I
> mirror distfiles? Have all dependent ports (on graphics/lib3ds,
> lang/expect, for example) suddenly became dead too?

You offer an "upstream" package and become the contact for issues. Just
mirroring bit-rotten crap isn't going to help anyone.  Unsupported
software is plainly and clearly a dead end, and I try to avoid running
in such dead ends whenever I can.

Possibly we should always mark ports for removal for three months after
the point in time when the maintainer gets reset to ports at .

> What has happened is users being unable to build perfectly working
> ports because of unneeded BROKEN's. FreeBSD ports are criticized for
> frequent build problems, there are talks about stable port branch
> for user to experience less frustration with ports tree, yet such
> plain sabotage is happening.

False. Users can set TRYBROKEN=yes and see how far they get.

False in that I don't see "FreeBSD ports that are criticized for
frequent build problems".

Also all past experiments about stable port branches concluded that they
just rot and are hence pointless.

But if you want stable ports, you can deinstall all FreeBSD ports and
install those from pkgsrc. It works, is supported on FreeBSD and
DragonflyBSD among lots of others, and it does quarterly "stable"
branches that pull up just security fixes, but it has a lot fewer
packages available and I find the tools inferior to ours.

But don't make false claims about the ports tree or proposals that
turned out to be non-workable for us.