Jeremy C. Reed wrote:
I added tech-security to this old thread. On Wed, 28 May 2008, Adrian Portelli wrote:The cvs log mentions an "upgrade script". What/where is that?Try http://cvsweb.netbsd.org/bsdweb.cgi/src/dist/bind/bind2netbsd ?This seems unnecessarily complex and seems to be missing several steps.
Worked well for me when I was using it for generating the patches for netbsd-4.
If I were you I'd diff the 9.3.2 and 9.3.5 releases from ISC, clean out all the RCS cruff, apply it to src/dist/bind and see what happens.Many conflicts over eight files.
Yeah, that would be right. I'd guess that these would be due to backported security patches and/or local patches that we have done that haven't been pushed up stream. I usually manually deal with the conflicts or in the case of the security patches I revert the local ones and then apply the diff from the updated version (which includes the security fix). This is just a fact of updates when we fall a few versions behind.
So instead I removed / fixed RCS ident. Copied files into place.
Ok, I haven't done it that way before but let's see how it goes :)
Here is a patch for testing. I built and I am running it on NetBSD/i386 3.1.http://reedmedia.net/~reed/tmp-98c74t3btgirfu/src-bind-patch-for-netbsd-3-for-bind-9.3.5-P1.diff.gz (1.2 MB)
Thanks very much for that I'll do some testing and we'll get it submitted as a pullup.
Also that has bogus files and windows source that can probably be removed.
AFAIK we typically don't remove anything before a cvs import is done into dist/. BTW what do you mean by 'bogus files' ?
Updating this is way too complex. Maybe using the scripts and instructions found in bind2netbsd and the files it references would be better. But they are missing steps (like should a "cvs up" be done between the steps in the different scripts?).
That I don't know. Like I said I've used these scripts before to generate patches and they have saved a lot of leg work for me.
adrian.