Re: Releasing individual files instead of whole directories

[Eric Siegerman]

On Wed, Sep 05, 2001 at 03:56:18AM -0400, Greg A. Woods wrote:
> [...] there's even some hint in the manual that "cvs release
> file" might be the equivalent of "cvs unedit",

Hmm, neither the documentation on "cvs release", nor its -H help,
mentions the possibility of applying it to individual files, and
in fact, in 1.11.1p1, "cvs release foo", where foo is an existing
CVS-tracked file in the current (sandbox) directory, gives the
error:
        cvs release: no such directory: foo

Do you mean this bit, from the documentation on "cvs watch add"?
>    [you receive `unedit' notification when]
>         Another user has applied the `cvs unedit' command (described
>         below) or the `cvs release' command to a file, or has deleted
>         the file and allowed `cvs update' to recreate it.

I take the "release" bit to mean "when another user has applied
`cvs release' to a directory containing the file in question".
In other words, "cvs release directory" clears any outstanding
"edit"s on the contained files -- which it must do, just as, at
the kernel level, exit() must close any open files, and that in
turn must clear any outstanding fcntl() locks.

> though even if it works that way I would suggest it's a bug and
> it shouldn't

Looks as though it doesn't.  If it did, I'd agree.

> "cvs release"
> should always only be thought of as the compliment of "cvs checkout")

Yup!

--

|  | /\
|-_|/  >   Eric Siegerman, Toronto, Ont.        address@hidden
|  |  /
With sufficient thrust, pigs fly just fine. However, this is not
necessarily a good idea.
        - RFC 1925 (quoting an unnamed source)