Subject: Re: Simultaneous builds
To: NetBSD Packages Technical Discussion List <>
From: Jim Wise <>
List: tech-pkg
Date: 01/16/2002 15:25:05
Hash: SHA1

On Wed, 16 Jan 2002, Greg A. Woods wrote:

>[ On Wednesday, January 16, 2002 at 13:34:20 (+0100), Alistair Crooks wrote: ]
>> Subject: Re: Simultaneous builds
>> Sorry, Greg, I don't have time to discuss this side issue right
>> now.  As I said, I'd be happy to see a well-discussed PR from you
>> on the subject of OBJMACHINE and OBJHOSTNAME, but mere contradiction
>> of anything I say isn't really the way to progress things.
>I'm sorry you cannot seem to follow what I thought was very detailed
>reasoning I gave for why OBJHOSTNAME (and OBJMACHINE) are totally
>unnecessary.  I do not know how I might express it in any better way for
>you to understand more easily.

If OBJHOSTNAME and OBJMACHINE are `totally unnecessary' _to_ _you_,
don't use them -- they're not even on by default, for crying out loud.

I find both of them _very_ useful in different contexts (including
OBJHOSTNAME, which I was _very_ happy to see added).  Last time I
checked, `is useful to gwoods' (or any other single user) is _not_ the
barometer for importing features into pkgsrc.

>I certainly will not be wasting my time on a formal PR until we can come
>to a more common understanding of the issues and requirements here in

Translation:  `I'm going to complain about 800 alleged bugs, but until
you agree with everything I say about the direction pkgsrc should take,
I'm not willing to take even the most trivial of steps to help see any
of those alleged bugs fixed.'

>> Oh, and you are mistaken - _HOSTNAME is added to MAKEFLAGS, so that
>> submakes don't re-evaluate it. You can verify this by looking at
>> the output of PKG_DEBUG_LEVEL=2 during a normal make.
>Ah, well that helps a bit at least -- meaning I won't have to rip it out
>in my local copy to maintain current performance levels.  Thanks for the

I think that's the core of the issue here.  In about the last dozen
threads that have come up on this list, your response to almost any
suggestion made is `well, that's not what I do in my private tree, so it
_must_ be a bad idea.'.

This isn't tech-gwoodssrc.  No really.  Everyone would love to hear
rational discussion of useful directions for pkgsrc, but telling us that
you don't want a feature so it's not important (as per concurrent
makes), that you don't implement a feature in your own patches, so it's
not important (as per OBJHOSTNAME, which I for one am _very_ happy
about), or constant poking at alleged bugs which you are for some reason
completely unwilling to file a PR for (as per too many examples to list
here) is just Not Useful.

- -- 
				Jim Wise
Version: GnuPG v1.0.6 (NetBSD)
Comment: For info see