Re: Making DESTDIR support mandatory

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

  Hello John,

jnemeth%victoria.tc.ca@localhost (John Nemeth) writes:

> On Aug 28,  7:58am, Aleksej Saushev wrote:
> } Joerg Sonnenberger <joerg%britannica.bec.de@localhost> writes:
> } > On Thu, Apr 07, 2011 at 01:11:35AM +0400, Aleksej Saushev wrote:
> } >> Thomas Klausner <wiz%NetBSD.org@localhost> writes:
> } >> 
> } >> > Since the branch is cut, perhaps we can progress on the destdir issue.
> } >> >
> } >> > Btw, thanks for the people who converted another few packages to
> } >> > destdir support!
> } >> >
> } >> > I suggest the attached patch -- this makes packages not build if they
> } >> > are not using destdir support, as incentive for people interested in
> } >> > them to fix them.
> } >> >
> } >> > We can then decide before the next branch if we want to revert the
> } >> > patch, just leave it in (my suggestion), or go further and actually
> } >> > remove the BROKEN packages.
> } >> >
> } >> > Comments?
> } >> 
> } >> This is plain wrong. Convert qmail and nagios to destdir first,
> } >> then we can return to this discussion. (Declaring working packages broken
> } >> isn't nice way to go. It isn't correct either.)
> } >
> } > Declaring them as broken is the only way of getting people interested in
> } > fixing them. Please read the time frame I mentioned in the original mail
> } > again.
> } 
> } You had the same time to fix qmail and nagios as well as any other person.
>
>      So did you.  And, given that qmail has a MAINTAINER why should he
> fix it?  BTW, I believe he has done a large number of conversions.

I don't deny it. Yet it is different to handle easy conversion where you
only check if it is supported by the package already, and when it involves
diving inside convoluted build systems like that of GHC or qmail.

> } Except you were working on this actively, while noone else understood
> } what the fuss was about, and you didn't care to explain. Now you have
>
>      Actually, I suspect that a fair number of people understand what
> the "fuss" is about.  If you don't then it is about time that you
> learned and/or took sometime to think about it.  If you think about it,
> it is fairly obvious:
>
> - clean binary packages with correct PLISTs
> - /usr/pkg managed completely by pkg_* tools without makefiles poking about
> - easier/cleaner development work when "installing" to a staging directory
> - etc.

I know it pretty well, and I spent a good deal of time in past explaining
what churn is going in pkgsrc with so many packages involved. I also
helped new packagers to understand what it is and how to achieve that.
Still I don't make illusions that pkgsrc's quality in public perception
suddenly improves by declaring some frequently used packages broken.

> } few hard cases remaining, you declare them broken. Perfect!
>
>      At some point you have to do that.
>
> } Why have you not converted them then? You had four years for it.
>
>      Why should he convert every package himself?

Why should working and frequently used packages be marked broken?

> } If declaring them as broken is the only way you can handle it, something
> } must be broken in your ways to deal with the world. Humans generally don't
> } need and don't like decisions made for them on grounds of some obscure cult.
>
>      It is hardly an obscure cult.
>
>      Instead of being a Luddite and making ad hominem attacks why don't
> you help with making forward progress.  Every package I
> import/maintain/touch I converted to DESTDIR a long time ago.

I don't maintain all packages I use, and so do many other people.
Many people don't maintain packages at all, this is neither good nor bad
per se.

> } Your time frame is irrelevant here. Those packages worked and continue
>
>      No it isn't.  If you don't set a hard deadline then it will never
> get done.

I have converted all Fortran packages alone with no hard deadline set,
consequently the past December has never happened.

> } to work for people who are interested in using them. Breaking them
> } because you follow the cult of User Destdir, isn't the proper way to
> } attract attention.
>
>      It is the only way to get the project completed.

See above.

You seem to do the same mistake as Joerg and Thomas do. There're other
people in community who actively use packages Joerg and Thomas didn't
convert. Like I said above, there're packages that were easy to convert,
and there're packages that were hard to handle. I did that work as well,
and I know what it is. You can convert a hundred of "nice" packages in
an evening, while some packages require so much involvement, that you
can't do it for months. Declaring them broken just because some volunteers
failed and other just didn't care, is plain wrong.


-- 
HE CE3OH...