Re: deprecate --fetch-cmd option of bootstrap script

[Joerg Sonnenberger <joerg%britannica.bec.de@localhost>]

On Sun, Jun 12, 2011 at 08:04:41PM +0200, Alistair Crooks wrote:
> On Sun, Jun 12, 2011 at 08:00:14PM +0200, Joerg Sonnenberger wrote:
> > On Sun, Jun 12, 2011 at 07:26:37PM +0200, Alistair Crooks wrote:
> > > the idea of the bootstrap script is for users to be able to run this
> > > once, and have everything set up for them so that they can just start
> > > using pkgsrc. the original bootstrap script was done that way for a
> > > reason.
> > > 
> > > so joerg's suggestion might have been well meant, but falls short of
> > > what is expected.
> > 
> > I disagree. The default should just work -- if there are still specific
> > reasons for chosing something else, ok. The only major reason I have
> > heard of for not wanting tnftp are:
> > (1) It requires ncurses on many systems, so doesn't build out of the
> > box.
> > 
> > (2) It doesn't support HTTPS.
> > 
> > For those platforms, fetch would be a saner default. The second part of
> > course only applies if OpenSSL can be found natively.
> 
> Yeah, I think you're missing something - the default does just work. 

No, it doesn't always work. In fact, this is the issue most likely to be
hit on Linux.

> But what we're talking about here is the ability to run the bootstrap
> script and have the results just work, whether or not the user has an
> aversion to tnftp or not.  We know from bug reports that there are
> users in that class.  So we have to cater for them, rather than just
> ignoring them.

So what next? An option to specify the location of packages? An option
to change the work area encoded into mk.conf? I think you are confusing
the bug reports...

Joerg