On Mon, Mar 07, 2005 at 07:34:54AM +0100, Ivan Popov wrote:
> The compilation and changes I applied are documented in
> /coda/konvalo.org/sw/pm/1/TOP/c/coda/V/cvs20050303/L/2/NOTES
> and corresponding
> /coda/konvalo.org/sw/pm/1/TOP/c/coda/V/cvs20050303/L/2/BUILD/patch.*
> (including your contributions, Stephen, thanks - mentioned in .../NOTES)
I browsed through your patchsets.
patch.fix.killing.realms
patch.no-daemonize
Should be fixed, you're not applying those anymore either.
patch.no-docbook2man
Already fixed using a bourne shell implementation of which, iterates
through $PATH, trying to find the executable.
patch.-lresolv
Just committed a fix for that, configure now searches for res_9_init
which seems to pull in the libresolv library on Darwin.
patch.uint16
Fixes a problem in the modular clog patch.
patch.vldb.h.uint
Applied.
patch.nokrb4
I don't know why configure picks up kerberos4. I could split the
with-crypto flag into separate -openssl -krb4 and -krb5 flags if you
want more control over which libraries get linked in.
patch.pioctl.h
I guess this should be an autoconf test, but since we already use
the platform check for DJGPP, I guess it can't hurt. Applied...
Looking at the FreeBSD patch, I'm starting to wonder why everyone
seems to be needing this. Is there some other header that might
the necessary ioctl bits?
patch.worker.h
Seems like the right thing, the finer details of how to mount /coda
is one of those things we can't reliably test with an autoconf macro.
Applied.
patch.cunlog-protection-hack
Not necessary. It seems like 'patch.fix.killing.realms' actually
was making some change which resulted in the cunlog breakage. As
you're not applying that patch, this one isn't necessary.
patch.ctime
Maybe we really just need to include the correct time.h or
sys/time.h header?
patch.no-lresolv
Seems to be against modular clog. configure should pull -lresolv
into $(LIBS) if the platform happens to need it.
patch.modular-clog
Not sure if I want to merge it just yet, although it looks like it
is getting some decent testing already.
btw. the openssl dependency isn't necessary, we only link against it for
the md5 and sha1 checksum functions. If the library isn't found we use
a generic C implementation (I believe the original reference
implementation). openssl could be providing some optimized assembly
version for your platform, but I don't think we use these functions
often enough that that will result in a measurable difference.
Jan
