Re: SSH URI draft

[nisse%lysator.liu.se@localhost (Niels Möller)]

suehring%braingia.org@localhost writes:

> So there's a path after the host part and that path represents a command 
> or file or path to repository, etc?

It's a path to the repository, which is implemented a directory in the
file system. It is passed as a command line argument to the process
started on the remote machine (and the actual command to use, hg
--server or whatever, is implicit, and I agree it's a good decision not
to encode it into the uri. Would make some sense as an optional
parameter, though).

> A parameter should be used for this instead otherwise it seems rather
> unclear what's supposed to happen with that path element outside of
> the context of the application-specific usage.

Is there any other reasonable interpretation of a path-element than as a
name in the filesystem hierarchy on the ssh server? That said, this
syntax was of course designed for use in a specific application. And I
don't know if you want the standard ssh uri's being designed to cover
the use case of identifying to a particular directory on a particular
server.

> As a side note, why wouldn't the scheme use a . to indicate a relative
> reference (as recommended RFC 3986)?

What would that look like?

  ssh://shell.gmplib.org/./hack/private-gmp-hack

for a path relative to my home directory, and

  ssh://shell.gmplib.org/home/hgfiles/gmp

for an absolute file name? Are there any example where this syntax is
used? To me, it seems counter intuitive, but that is mostly because I'm
used to "/./" being equivalent to just "/"; I'm not familiar with
RFC3986 conventions for anchoring the path.

(And I'm just a mercurial user. I find its use of ssh uri's reasonable,
but I wasn't involved in any way when the used syntax was discussed or
decided).

Regards,
/Niels

-- 
Niels Möller. PGP-encrypted email is preferred. Keyid C0B98E26.
Internet email is subject to wholesale government surveillance.