Re: cvs has problems using LockDir when CVSROOT holds a symlink

[Mark D. Baushke]

Larry Jones <lawrence.jones@eds.com> writes:

> Mark D. Baushke writes:
> > 
> > The 'make check' step is still running, but here is the patch in its
> > current form.
> 
> Looks good, except I'd suggest moving expand_directory() to filesubr.c;
> I can envision more general use for it in the future.  

Hmmm... that means adding it to all of the filesubr.c versions:

  emx/filesubr.c
  os2/filesubr.c
  src/filesubr.c
  vms/filesubr.c
  windows-NT/filesubr.c

Is there any more generic place it could live instead rather than
duplicating all of that code?
  
> I also have a question: In parse_cvsroot(), why do you check for
> dirpath being NULL before setting it?

Your are right, it is not necessary. I will change the check to use:

    if (newroot->method == local_method)
        newroot->dirpath = expand_directory (newroot->directory);

Is that okay?

> It looks to me like that condition is always true. 

I'd hope so.

> Also, if method isn't local_method we should probably skip the call to
> expand_directory() and just dup the original string since only the
> server can meaningfully expand the directory name.

> On second thought, we should just leave it NULL instead of duping -- the
> client shouldn't be accessing that field anyway.

Okay. parse_cvsroot() will only call expand_directory() if it is the
local_method and local_cvsroot() will always call it.

        Thanks,
        -- Mark