tech-pkg archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Making DESTDIR support mandatory



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

> On Wed, Apr 06, 2011 at 03:15:05PM -0400, Greg Troxel wrote:
>> Right now, using DESTDIR is not the default.   I certainly agree that
>> this should change.
>> 
>> Your patch seems to be doing things in wrong order.  Specifically, I
>> don't understand why it is reasonable to mark packages broken because of
>> lack of DESTDIR support when the default is not to use DESTDIR support.
>> I just checked - at least on the 2011Q4 branch, 'make replace' on a
>> package uses DESTDIR with PKG_DEVELOPER=yes, and doesn't otherwise.  (I
>> have PKG_DEVELOPER=yes on almost all machines.)
>> 
>> Steps that I think should happen are:
>> 
>>   1) Change default to use DESTDIR, with or without PKG_DEVELOPER.
>>   Start having the norm for bulk builds to use this (perhaps they do,
>>   but I'd expect default behavior for public bulk builds).  Have an
>>   option to set it to the old way.  This is really the most important
>>   change, and I see no reason not to do this right now.
>> 
>>   This is is easy; just remove the .if on PKG_DEVELOPER on line 428 of
>>   bsd.prefs.mk, and replace with "USE_DESTDIR?= yes".
>> 
>>   2) Change the current warning to be always enabled.  I see no reason
>>   not to do this immediately
>
> I've taken these two steps, adapting your patch slightly. Thanks for
> providing it.

I am trying to be useful :-) and you're quite welcome - that sounds
good.

>>   3) Change the code, similarly to your patch, to set BROKEN if both a)
>>   USE_DESTDIR=yes and b) the package is not DESTDIR-ready.  I see no
>>   reason not to do this right now.
>
> I suggest the attached patch for this. Seems to work as expected for
> me.

While I said I saw no reason to wait, particularly after reading
Aleksej's comments, I think we should wait for fallout from steps 1/2.
We don't have a handle on the unintended consequences.  (If I knew of
any, I would have said, but there's what we don't know.)

>>   4) Remove the option to use other than DESTDIR mode.  Given the
>>   progress, I expect pretty soon the number of deficient packages will
>>   be really small, instead of just small.  But I don't think we've
>>   reached the time to do this step.
>
> Ok, so let's talk about this step again in a month or so.

We can of course discuss at any time, but I think the key is getting few
enough packages non-DESTDIR-capable that we think losing those is ok.

At some point the burden will shift from

  those who want to change to DESTDIR-oonly

to

  those who care about the last N packages, which are obviously fringe

and I don't think we're there yet.

Attachment: pgpyQCvmBPYb5.pgp
Description: PGP signature



Home | Main Index | Thread Index | Old Index