Subject: Re: Rework of the fetch target (was: CVS commit: pkgsrc/mk)
To: None <tech-pkg@netbsd.org>
From: Bernd Ernesti <netbsd@arresum.inka.de>
List: tech-pkg
Date: 02/01/2002 21:08:48
On Fri, Feb 01, 2002 at 05:53:22AM -0600, Frederick Bruckman wrote:
> On Fri, 1 Feb 2002, Bernd Ernesti wrote:
> 
> > On Sun, Jan 27, 2002 at 01:13:32PM -0600, Frederick Bruckman wrote:
> > > On Sun, 27 Jan 2002, Frederick Bruckman wrote:
> > >
> > I think you don't know what we get before from a make fetch-list.
> >
> > 1. we would only get an output, if the distfile for this package doesn't exists
> >    on the local system
> 
> Doh! I was wrong to defer the check until the script is executed.
> 
> > 2. only the distfile for this package would be displayed, not other distfiles
> >    from other packages, even if this package has dependencies.
> 
> I don't remember it that way. For some time, there's been a fetch-list
> target, and a fetch-list-one-pkg. fetch-list-one-pkg almost does what
> you're requesting, but it still (improperly) defers the check.

Hmmm, fetch-list-one-pkg doesn't work when you start it in pkgsrc, it only
works if you are in pkgsrc/xx/yy.

But this is no real problem if the output is shorter or non existent.

> > 3. you would not only get the distfile, but also get a command line which you
> >    could use to fetch this file.
> > but now you get the output even if the distfile is there and the output is now
> > unreadable due too the length of it.
> >
> > #2 and #3 is what I really want. That way I could fetch this distfile on a
> > different system, which has a faster connection, without having the pkgsrc
> > tree on this system.
> 
> OK. I have some ideas to make fetch-list much nicer. Give me a day or
> two to hammer out the details.

Thats ok, I have enough other things to do.

Bernd