[EMAIL PROTECTED] on 02/16/2000 06:04:11 PM
>[ On Wednesday, February 16, 2000 at 12:43:31 (-0500), Noel L Yap wrote: ]
>> Subject: Re: removing the need for "cvs add file" to contact the server....
>>
>> "cvs add dir" *must* operate on the local directory in so far as to create
CVS
>> admin subdirectories within those local directories. It should bear no
>> difference whether or not these local directories have files within their
>> hierarchies.
>
>Only in the case where getcwd() is an as yet unknown-directory and the
>command-line argument is another sub-directory.
>
>Note though that these three commands are supposed to be exactly
>identical in CVS:
>
> cvs wdcmd sub-dir
> cd sub-dir && cvs wdcmd "."
> cd sub-dir && cvs wdcmd
>
>(where "wdcmd" is any CVS sub-command that operates in a CVS managed
>workspace, including "update", "status", "log", "diff", "commit", "rm",
>and so on)
Exactly. Note also that sub-dir does not in any way differentiate between
having files somewhere within its hierarchy or not.
>> So then directory hierarchies containing zero files are also part of the
working
>> directory.
>
>Yes in so far as they are within the working directory. "cvs update -P"
>will delete them though if they are indeed truly empty, and of course
>since '-P' is supposed to be the default, and is forced to be the
>default when using branches, one must assume this will happen
>regardless. If you don't want CVS to delete an empty directory then
>make sure it's not empty! ;-)
This is fine by me. Again, note that the above doesn't mention "cvs add
empty-hier" at all.
>> They also shouldn't be ignored by "cvs add".
>
>They won't be purely ignored -- but nothing can happen unless there are
>un-ignored files within to be operated on.
One more time: it should be "cvs co", "cvs export", and "cvs up" that handle
this, not "cvs add" or "cvs ci".
Noel
