Ports completely and permanently hosed

[b. f.]

On 8/7/09, Erich Dollansky <erich at apsara.com.sg> wrote:
> Hi,
>
> On 07 August 2009 am 08:44:44 b. f. wrote:
>> Erich Dollansky wrote:
>> >I think that you hit the weakest point of FreeBSD. When a
>> > version number of a base port changes, hundreds or even
>> > thousands of ports have to be recompiled. It is basically the
>> > same effect as when the major version number of FreeBSD
>> > changes.
>>
>> The same is true of almost any build-from-source distribution
>> that uses shared libraries, not just FreeBSD.
>>
> of course, this is plain logic.

It may be, but a large number of people that complain on this list
seem unable to understand this, and speak of these problems as if they
are specific to FreeBSD or to Ports.

>
>> >If this would be synchronised with the main FreeBSD releases,
>> > it would have a minor effect on users.
>>
>> But please don't attempt to slow needed development by making
>> *(&@Q%#%@!!!! suggestions like this.  If you need a seat-belt,
>> put it on -- but don't wrap it around everyone's neck.
>>
> So, why is there a ports freeze just before a new release?
>
> Isn't it done just out of the same reason?
>
> They want to have a stable ports tree on the day of the release.

Yes, and for building a stable subset of packages beforehand to ship
with the release.  But these freezes are considered to be a necessary
evil, to be removed as soon as possible, and not something that should
be in place from release to release. The current version of the Ports
tree is supposed to be the leading edge of (downstream) development.

>
> As I said, my seatbelt is the freeze whenever a basic library gets
> changed.

Okay then, but why should you want to impose this on other people who
aren't as conservative, or who need more frequent revisions?


b.