Re: libiberty/partition.c

[Ian Lance Taylor <ian at zembu dot com>]

   Date: Fri, 24 Mar 2000 19:52:32 -0700
   From: Jeffrey A Law <law@cygnus.com>

     In message <20000325013507.4371.qmail@daffy.airs.com>you write:
     >    Date: Fri, 24 Mar 2000 16:30:58 -0800
     >    From: Eric Christopher <echristo@cygnus.com>
     > 
     >    Doesn't appear to exist in the tree.  Looks like this change:
     > 
     >    2000-03-09  Alex Samuel  <samuel@codesourcery.com>
     > 
     > 	   * Makefile.in (CFILES): Add partition.c.
     > 	   (REQUIRED_OFILES): Add partition.o.
     > 	   (partition.o): New rule.
     > 	   * partition.c: New file.
     > 
     >    Didn't make it in - though it made it into egcs.  Wonky.
     > 
     > The binutils/gdb repository and the egcs repository are separate.
     > Checking something into egcs does not imply checking it into
     > binutils.  However, if something is approved for checkin to the egcs
     > version of libiberty, please feel free to bring it over to the
     > binutils version of libiberty.
   I've been wondering how well CVS would work if there was a symlink for the
   libiberty & include directories until we can merge the sourceware & gcc
   repos.  I haven't done any experiments, though I'm rather leery given the
   problems the FSF had with symlinks in the CVS tree.

I just looked at the CVS code a bit, and I doubt that it would work
correctly.  The find_dirs call appears to explicitly skip symlinks
when searching for directories.

As I recall, the FSF was symlinking at the file level, not the
directory level.  I would be dubious about that too, because of the
PreservePermissions support which, if enabled, causes CVS to version
control symlinks.

Ian