Subject: Re: calling package INSTALL and REQ (REQUIRE!) scripts from bsd.pkg.mk
To: NetBSD Packages Technical Discussion List <email@example.com>
From: Greg A. Woods <firstname.lastname@example.org>
Date: 03/16/2000 01:25:06
[ On Wednesday, March 15, 2000 at 23:33:43 (+0100), Hubert Feyrer wrote: ]
> Subject: Re: calling package INSTALL and REQ (REQUIRE!) scripts from bsd.pkg.mk
> On Wed, 15 Mar 2000, Greg A. Woods wrote:
> > > this should be done via pkg/INSTALL (pre-install), so it gets checked for
> > > binary packages as well. (I dunno, though, if that file is run from
> > > bsd.pkg.mk - please check)
> > No, pkg/INSTALL is not run from bsd.pkg.mk. But it should be.
> "depends". Very often, the software's Makefile already does some of the
> things in the INSTALL file.
Hmmm... Yes, I suppose that's very often possible. However in that
case I would hope it's possible to write the INSTALL script in such a
way that either the re-doing is harmless, or that the "done-ness" can be
detected and duplicate efforts thus avoided....
That said though I've not yet encountered such a situation in any of the
packages I've worked on (very few out of the total these days, I admit).
I.e. I'm calling into question your claim of "very often". Is there any
scientific guesstimate as to how many existing INSTALL scripts might
have to be at least examined and tested to see if they can indeed be
safely run regardless of whether the install is from pkgsrc or a binary?
I only count 26 INSTALL scripts in total right now so even if they all
need careful examination it doesn't sound like even a large effort....
Of course given my experience with even a dozen dozen (144) packages I
can guess that there should be at least twice, if not three times, that
many INSTALL and/or REQUIRE scripts for fully correct binary installs.
Greg A. Woods
+1 416 218-0098 VE3TCP <email@example.com> <robohack!woods>
Planix, Inc. <firstname.lastname@example.org>; Secrets of the Weird <email@example.com>