[PATCH] Added NoDelay config option and nodelay subsystem option

mouring mouring at etoh.eviladmin.org
Sat Feb 2 02:15:08 EST 2002


> On Fri, 1 Feb 2002, Markus Friedl wrote:
> 
> > On Fri, Feb 01, 2002 at 11:38:02AM +0100, Tobias Ringstrom wrote:
> > > IMHO, you
> > > would also benefit from a development model with a development and a
> > > stable branch.  
> > 
> > I don't see why?
> > 
> > The OpenSSH-current from the CVS is the development branch.
> > I don't see how this helps for testing your patches. I don't
> > think candidate patches should go into the development
> > before everyone tests the patches. Even patches for
> > the development branch need testing before the get into
> > the development branch.
> 
> Testing yes, but how much?  I think mailing list members are much more
> likely to test release candidates (or even snapshots) than to patch and 
> test individual patches.
> 

http://www.openssh.com/portable.html  has snapshots every night.  Yet
no large mass of people are testing it.

> Basic testing by the developer and a quick review should be enough to 
> apply the patch, IMHO.  If further user based testing reveals problems, 
> they can be fixed, or the patch reverted.
> 
> > However, we're going to have a time based release schedule,
> > OpenBSD benefits from this, and OpenSSH will probaly, too.
> 
> Just add release candidates some time before the release, and I'm sure it 
> will work out nicely.
> 
There is a at least 3 calls I try to make before we release portable tree
asking for people to do compile tests of the current snapshots.. Some
days I hear feedback.  Some days I don't.  Without flying out to every
person's house on this list and beeting them with a blunt object until
they finally download and test there is not much we can do.  (Personal
Note:  Feel safe.. I'm not a person to do such things.  =)

I always try to get a Linux (Mandrake is this flavor of the month), 
OpenStep 4.2, and Solaris 2.5.1 and 7 compiling in before we make a 
release.  I know a few others in the portable group try to test on 
all SCOs before the relase, but there are too many possible combinations 
to test and I know for a fact I'm not in the position to test everything 
from krb4 to s/key at my house (Even if I've compiled s/key in the past to
help debug why it was broken).

Doing OpenSSH-3.1.0-Pre2, pre3, pre4-mouring-fixing-bug  release will
not help us.  Hell it barely works in the Linux community!!! =)

On a side note:  ANY PATCHES THAT ARE NOT IN BUGZILLA WILL/MAY BE LOST.  
PLEASE ENSURE THEY GET INTO THE TRACKING SYSTEM.  

- Ben



More information about the openssh-unix-dev mailing list