[quagga-dev 5847] Re: Update: unnumbered PtP support for OSPF.

Joakim Tjernlund Joakim.Tjernlund at transmode.se
Mon Sep 1 18:25:07 BST 2008



> -----Original Message-----
> From: paul at clubi.ie [mailto:paul at clubi.ie]
> Sent: den 1 september 2008 18:18
> To: Joakim Tjernlund
> Cc: quagga-dev at lists.quagga.net
> Subject: RE: [quagga-dev 5832] Update: unnumbered PtP support for OSPF.
> 
> On Mon, 1 Sep 2008, Joakim Tjernlund wrote:
> 
> > If you had, I would have started my new tree from there so it that
> > is not a problem.
> 
> Sure, but you can't know what I've done locally.
> 
> > Also, I haven't asked you to pull anything yet.
> 
> Ah.. Ok. You don't expect me to pull except on explicit request. See,
> there's an assumption ;)

Yes, and an important one too. You should not assume that stuff is ready
to be pulled because there is a tree with stuff in it. :)

> 
> > I sent a few patches that needs to go in first and which I think is
> > safe. I sent those first few because I have a hard time to get
> > anything into the tree.
> 
> > I know I don't have a history with Quagga so you obviously
> > need to look extra on my patches, but this is getting a bit
> > annoying as not even the simplest SEGV bug fix has gone in or
> > commented upon.
> 
> Patience young grass hopper. :)

hehe, soon running out though :) The first submission for the SEGV
problem was 2/7, how much patience do I need?

> 
> Thanks for seperating out the obvious stuff. I can see you've pulled
> the obvious fixes back into your own unnumbered tree, so you're not
> blocked on these getting into 'master'.
> 
> >> We need to work out some kind of protocol here..
> 
> > Yeah, for now I am happy with either patches or if you
> > cherry-pick/rebase your self. That way Quagga will still have a
> > nice linear history.
> 
> Well, let's be clear: I don't think it's possible to keep a linear
> history. We could try fake such a history, but why go to the bother?
> If some work gets done alongside the 'master' branch before
> eventually being included, I don't see a problem with the history of
> 'master' reflecting that.

True, but at the same time you don't want garbage into the tree either?
Assume I had publish my first attempts and the just kept fixing stuff as I
go along and then asked you to pull that?
So I think the personal trees must be allowed to rebase or be deleted
at the authors will. When ready you get an 
"Please pull branch xxx to receive foo" or a patch series.

> 
> > However my repo contains extra stuff that isn't ready for general
> > consumption (The auto-unnumber command, not sure this is a good
> > idea yet) so if you feel ready to integrate Unnumbered support, I
> > can prepare a patch series with only relevant stuff in it.
> 
> Well, how well tested is the unnumbered stuff at the moment?

Well, I am happy with the testing I have done, but I haven't got
much feedback from others. Typically I just get to hear when there
is a problem, usually this is due to misunderstandings, trying to back port
to stable Quagga or the run stuff on a 2.2 kernel. Then things become quiet :(

> 
> E.g. I'm not in a position to test ospfd very well at the moment, so
> I'm really curious to hear from people trying out your work first.

Me too!

> (Has anyone ever looked at using tun/tap and some kind of virtual
> switch for testing ospfd? Anyone know of generic tun/tap switches?).
> 
> In case you think I'm picking on you, I and other maintainers have
> maintained patches (i.e. branches) outside of CVS HEAD for prolonged
> periods of time (indeed, I've got patches I use, which I never got to
> put back). That kind of thing is perfectly normal, particularly with
> more non-obvious patches.

Sure, as have I, but this is a bit different. I have done my best to work
out a solution which is acceptable to everyone with the explicit intention
to get it upstream. This is a missing core function that has been requested by
several people so I am a bit surprised that this is progressing so slow.

 Jocke




More information about the Quagga-dev mailing list