[Cvs-dev] Re: cvs-chpass.patch (former: cvs-passwd.patch)

[Prasad J Pandit]

  Hello Derek! :)

On Mon, 26 Jun 2006, Derek R. Price wrote:
> How about this.  Since you have implemented something slightly different
> than CVSNT:
>
> 1.  Call your command `passwd', as you did originally.

  Great! (I'm happy and dancing)  :) :) :)

> 2.  In client mode, accept as options all of the options CVSNT accepts
> in addition to your new options, and forward them to the server,
> allowing the server to reject them if it wishes.  These are (from
> <http://www.cvsnt.org/manual/html/passwd.html>): -a, -x, -X, -r
> SYS_USER, -R, and -D DOMAIN.  The complete CVSNT syntax is:
>
>         cvs passwd [-axXR] [-r SYS_USER] [-D DOMAIN] [USERNAME]

  Okay!

But why let server reject the (invalid)options, shouldn't that be done at 
client side? Or

It'll be more clear, if you could give me some case(s) when server rejects 
the (invalid)options??


> 3.  In local and server mode, accept only the options you wish to
> implement, including any options not implemented by CVSNT, but if you
> have implemented any similar functionality, try to merge it under the
> same syntax/options as CVSNT.

 Ummn... I didn't get this.

> 4.  Use the same client/server protocol for passwd as was defined by
> CVSNT.  Without looking at your patch, I would guess this shouldn't
> change your implementation much.  Glancing at the CVSNT client
> implementation, it looks like it is just using send_arg for the
> password, and sending "*" for USERNAME when it isn't specified (treating
> the command as if the syntax was:
>
>         cvs passwd [-axXR] [-r SYS_USER] [-D DOMAIN] USERNAME [NEWPASS]
>
> and a username of "*" means, "the current user").

 Fine! I'll do the changes and get back as soon as possible.

Thank you!
--
regards
   -Prasad
PS: Please don't send me html/attachment/Fwd mails