Re: Problem with add directory using NT cvs server

> > 3. Added a directory to the root.              [FAILURE]
> >    The message provided by the server reads:
> >    " cvs [server aborted]: cannot add to
> >      D:/cvsroot/nosdev/CVSROOT/Emptydir "
> >
>
> That is correct behaviour IMHO.
>

Hmmm... it should be possible to add a module, shouldn't it?
Or am I missing a standard CVS feature? Do I always have to
"import" a module (from an empty directory)? Say, like:

1. mkdir 123
2. cd 123
3. cvs import -m"..." 123 vendor tag

Why not:

1. mkdir 123
2. cvs add -m"..." 123

> can you please describe the debug example some more? [...]

Si.

SERVER configuration

 - prefix to all repositories is "D:/cvsroot"
 - one repository is called "/nosdev"
 - there-in are some modules "CVSROOT", "nos", docs", ...

CLIENT setup

 - Sandbox is "E:\projects\nosdev\calix"
 - Content of the _magic_ (and hand-edited) "CVS/" folder:
    Entries   : "\CVSROOT\\\\ \newline \nos \newline ..."
    Repository: "CVSROOT\Emptydir" (or "." - it depends ... :)
    Root      : ":pserver:sormuras@calix:/nosdev"

> that case is catched by the removal of initial dot, but
> in case the dot is in between there's a problem...

It was quite interesting to find the place_s_, where the
dot and other prefixed strings are removed. I first thougt,
that only the StandardAdminHandler does the job, but the
Client is the real worker here. Maybe, we should do it only
in one place.

Christian