Subject: Re: CVS commit: pkgsrc
To: Hubert Feyrer <>
From: Adam C. Migus <>
List: tech-pkg
Date: 01/03/2004 16:33:20
Hubert Feyrer said:
> On Sat, 3 Jan 2004, Jeremy C. Reed wrote:
>> I don't understand what you mean about "PKGREVISION for something
>> that it
>> wasn't made for". What I did was normal and has been done several
>> times
>> before.
> PKGREVISION is a counter that tells (roughly) the number
> user-visible
> changes to a pkg, when the distfile was not changes.
> Using it for something else is a gross hack, and if there's a need
> to
> express something else, go and add a new handle to do such things,
> and
> don't overload existing interfaces.
> The second sentense quoted above seems like a bad excuse to me,
> sorry.
>  - Hubert
> --
> Hubert Feyrer <>

The effect of changing PKGREVISION is, in essence causing a "new
build."  Correct me if I'm wrong but when I see "nb{x}" I always
think "nb" stands for "new build."  Thus regardless of what
PKGREVISION intuitively means to someone reading the Makefile, what
it does is cause a new build.  As Jeremy points out this has the
effect of updating binary dependencies and satisfying the general
requirement of a package changing without its underlying
distribution changing.

The name PKGREVISION does indeed imply the meaning discussed;
changes to the package itself.  Such changes however have little
affect on anything in this context making the variable more less a
dummy value if it were not satisfying the need of triggering a "new
build."  For what it's worth if I want to see if a package changed I
read the CVS logs.

So the way I see it you're basically proposing to rename the
PKGREVISION variable to something appropriate and create a
PKGREVISION variable that reflects what you think PKGREVISION should
reflect when the latter is arguably redundant though I recognize not
everyone wants to visit cvsweb.