Re: [HEADS UP] packages not supporting destdir

[Aleksej Saushev <asau%inbox.ru@localhost>]

Thomas Klausner <wiz%NetBSD.org@localhost> writes:

> On Sun, Apr 01, 2012 at 05:01:52PM +0400, Aleksej Saushev wrote:
>> > I plan to remove these packages after the freeze if noone converts
>> > them. There was plenty of time.
>> 
>> No, there wasn't, and there's still no need to remove them until it is
>> proved that these packages don't build and don't install for a long time.
>
> I've been asking for these and other packages to be fixed in particular for 
> over 10 months now.
> http://mail-index.netbsd.org/pkgsrc-users/2011/05/18/msg014430.html]
>
> What is enough time?

Until it either gets fixed or dies out naturally.

This issue is serious. We do harm to NetBSD and pkgsrc acceptance by removing
packages on arbitrary grounds. You should present strong reasons why the lack
of staged installation support should be the reason to remove fully functional
packages. Note that it is fundamentally different from removal of old bulk
build code or, perhaps, alternatives framework. We have pkg_comp and other
facilities to isolate package builds, if potential side effects do really 
matter.
There's absolutely no reason to remove packages that don't support staged
installation, especially when it is done just for mere convenience of few 
developers.

There're other issues, communicational and organizational. I think that
forcing some developers to work on packages just to appease some
strange tastes of others is plain wrong. If you have time and consider
it really important, why don't you do it yourself? O'Caml and Haskell
are still popular programming languages, and you cannot change that,
Courier IMAP is still used, and you cannot change that. I don't
understand why you extort my time to "fix" fully functional packages
so that I and some other people could continue using them. To me it
looks exactly like that. I and many other people don't have your
prejudices on building some packages as superuser, or on whether the
number of broken packages is 300 or 200 given the overall number of
packages or that the number of broken packages on Solaris is around 5000.

In addition, if you really want to clean the code up a bit, you have
enough to do. E.g. old bulk build code. Why is it still not removed?
Contrary to packages this code is purely internal framework, not visible
to anyone, pbulk superseded it long ago. Similar reasoning applies to
alternatives. If the framework is not used, why is it still there?
I'm sure that you can find other things to clean up. Why are you pushing
removal of user-visible things rather than purely internal code instead?


-- 
HE CE3OH...