Re: Feature Request: admin files for "cvs import" and "cvs add<dir>"

[Wu Yongwei]

Alexandre Augusto Drummond Barroso wrote:
> I think the discution is limited to what can be done about the import
> issue, because cvs doesn't deal with the concept of "adding directories"

A user could add a directory without triggering the script in commitinfo,
which I regard as a potential security hole.  No matter what the concept is
in CVS.

> or "removing directories" directly. AFAIK, what it really does is "remove
> all files from certain directory" to remove a directory or "add files that
> have a common path witch doesn't exist before" to add a directory.

I doubt it.  When I type "cvs add test" and test is a directory, the test
directory (as well as a log message) will be generated on the server.  The
only problem is that the script in commitinfo is not triggered.  But I see
it really a hidden commit operation.

>
> Zandall.

However, I agree that this issue is not as important as the "import" issue.
In most circumstances it is harmless and recoverable.  But why not prevent
it beforehand?  In some cases you cannot expect that every developer is
smart and knows how to do things correctly.

Best regards,

Wu Yongwei