RE: cvs version w/o repository.

[$B8q9>1QN4(B]

From: "Cameron, Steve" <[EMAIL PROTECTED]>
Subject: RE: cvs version w/o repository.
Date: Sat, 29 Jul 2000 03:01:31 +0900 (JST)

 | especially useful if you have no knowledge of anything about the remote
 | side, (e.g. it's anonymous pserver access).

In trobleshooting of anonymous access with pserver method,
a client cannot invoke `version' before authenticated.
If bug in authentication phase, no user report the server version.
This is caused by the protocol layer;
pserver-handshake underlies C/S protocol, you know.

I dont propose changes, KISS.
If cvs C/S protocol stack would be changed,
RQ_AUTHENTICATED could add.

--
KOIE Hidetaka <[EMAIL PROTECTED]>

Re: cvs version w/o repository.

[Larry Jones]

Cameron, Steve writes:
>
>   [smc]  Wasn't "cvs version" recently introduced precisely so that
> you could
>   see what version of CVS the _remote_ repository was running?

Yes, Hide-san's point was that you shouldn't have to specify a valid
root directory on the server just to find out what version it is.  It's
a valid enough point -- I implemented it so that the server is happy to
respond to a version request without a preceding Root request (although
I neglected to document it that way), but I didn't implement it that way
in the client for a variety of reasons.

-Larry Jones

Life's a lot more fun when you're not responsible for your actions. -- Calvin

RE: cvs version w/o repository.

[Cameron, Steve]

I wrote: [...] 
>   To see what version of CVS you're running locally, use "cvs -V"
[smc]  I meant "cvs -v"

RE: cvs version w/o repository.

[Cameron, Steve]

> From: [EMAIL PROTECTED] (Larry Jones)
> Subject: Re: cvs version w/o repository.
> Date: Sat, 29 Jul 2000 00:18:32 +0900 (JST)
> 
>   | > cvs version command may not access a repository.
>   | 
>   | Why not?  It doesn't seem like an unreasonable restriction to me.
[smc]  Wasn't "cvs version" recently introduced precisely so that
you could
see what version of CVS the _remote_ repository was running?
To see what version of CVS you're running locally, use "cvs -V"

This was added to help with debugging why client server connections
might fail, so for example you might be able to tell if the remote
side
of a connection might be running a buggy version of CVS or
something,
especially useful if you have no knowledge of anything about the
remote
side, (e.g. it's anonymous pserver access).

RE: cvs version w/o repository.

[$B8q9>1QN4(B]

From: [EMAIL PROTECTED] (Larry Jones)
Subject: Re: cvs version w/o repository.
Date: Sat, 29 Jul 2000 00:18:32 +0900 (JST)

  | > cvs version command may not access a repository.
  | 
  | Why not?  It doesn't seem like an unreasonable restriction to me.

Beause:
`cvs version' displays it's version, not a configuration of repository.

This is a weak reason, I have no unyielding intention.

Proposed amendment:
`cvs version' is casted away, `cvs introspect' is newly introduced.
 "cvs intro -v" will display the version and an infomation related
to the repository if any.  In addition, some notice like
 "cvs --version" will be displayed.
I prefer "cvs config" also.  `cvs version' is a synonym of it.

--
KOIE Hidetaka <[EMAIL PROTECTED]>

Re: cvs version w/o repository.

[Larry Jones]

KOIE Hidetaka writes:
> 
> cvs version command may not access a repository.

Why not?  It doesn't seem like an unreasonable restriction to me.

> "valid-requests" is used instead of "noop",
> because "noop" command is not RQ_ROOTLESS.

It would make more sense to change noop to be rootless.

-Larry Jones

We seem to be out of gun powder. -- Calvin