Re: Symlinks and CVSROOT

[Larry Jones]

Eric Siegerman writes:
> 
> Not to cast doubt on your comments, Larry -- you certainly know
> the code a lot better than I do -- but rather to understand more
> deeply:
>   - What's the problem with this?

The underlying problem is that CVS manipulates path names and moves
around in directory trees.  It knows that there may be multiple ways to
name the same file, so it tries to make sure that the path names it
creates are based on the actual path the user specified and not some
synonym that may not be obvious to the user should it appear in an error
message, but not all parts of the code are sufficiently concientious.

>   - Why might it work in some situations but not in others?

Because some of the code does things right, but not all of it.

>   - I find this technique quite useful (obviously).  Would it be
>     feasible to change CVS to make it as dependable for everyone
>     as it seems to be for me?  (N.B.: I'm simply asking for an
>     opinion, not for you to do the work.)

Of course it would.  A number of people have even submitted patches from
time to time, but no one has ever (to my knowledge) completely analyzed
the problem (what works, what doesn't), fixed it in accordance with the
above guidelines, and provided test cases to verify the fixes.

-Larry Jones

I told her to expect you to deny everything. -- Calvin