|
|||||||||||
|
Re: linux-ipsec: Latest snapshot and 2.0.35
From: Henry Spencer <henry(at)spsystems.net>
Date: Tue Jul 28 1998 - 20:56:15 EDT
Correct. We noticed this a while back.
> I've got mixed emotions about using that symlink idea.
Frankly, so do we. If you've been following the history of the way "make insert" works, you'll see that we've tried more than one approach. The real problem is, quite simply, that there is no right way to do it, only wrong ones. The trick is finding the one that's least wrong. I'm not convinced we're there yet, but at the moment all the alternatives look worse. > ...If you're going to modify files in the kernel source tree,
Trouble is, then you have two copies of the code, with all the problems that can ensue with them getting out of step. It also makes it hard for the kernel parts to refer to non-kernel parts -- for example, to the libraries that the two share. Sounds likely. At the very least, the protocol.c changes need to be wrapped in "#ifdef ipsec-is-not-a-module". While we consider static linking preferable for security-sensitive code, it *is* our intent that it continue to work as a module.
Henry Spencer
henry@spsystems.net
(henry@zoo.toronto.edu)
Received on Wed Jul 29 00:10:29 1998This archive was generated by hypermail 2.1.8 : Wed Aug 23 2006 - 12:59:25 EDT |
||||||||||
|
|||||||||||