[distcc] Re: masquerade for 1.2.3

[Martin Pool]

On  4 Mar 2003, Wayne Davison <wayned at users.sourceforge.net> wrote:
> On Wed, Mar 05, 2003 at 08:15:25AM +1100, Martin Pool wrote:
> > Wayne, could you write a few paragraphs for the manual about this?
> > Either plain text or a diff to the SGML is fine.
> 
> Sure, I'd be glad to.
> 
> On a somewhat related note, I have been meaning to ask you about the
> "/MAX" syntax since I was thinking about helping to document it.
> However, I'd need to know if you've yet decided to keep the syntax or
> not, if you'd like to me submit such a patch (or prefer to do it
> yourself), and what the preferred syntax is for combining it with a
> port:
> 
> I had thought that the full syntax would be HOST:PORT/MAX (since I think
> of the HOST:PORT as a set that specifies a unique distcc resource).

That was what I originally planned.  Rusty's patch used HOST/MAX:PORT
and I hadn't decided whether to change it or not.

> However, it appears that the parser expects this to be HOST/MAX:PORT.
> Do you have a preference for either format?  I think that the former is
> slightly better, and would be glad to tweak the code to accept it.

I agree that the other one is a bit more sensible.

It kind of depends on what the ssh syntax is.  It seems to me that a
nice way to write it is

  @HOST
  USER at HOST
  USER at HOST/MAX

although something like this might also make sense, even it's a bit
more verbose 

  ssh:USER at HOST/MAX
  ssh:HOST

I had wondered about allowing the name of the remote program to be
specified, by 

  @otherbox:distccd-1.2
  @otherbox:/usr/local/bin/distccd-1.2

This sort of thing seemed to sometimes be useful for rsync.  Obviously
having a path in there causes problems with adding a /MAX at the end.
This may have been what Rusty had in mind when he wrote his patch.

I'm not sure specifying the remote command per host is really terribly
useful though.

-- 
Martin