[quagga-dev 10863] Re: [PATCH 3/4] ospfd: ospfd-unnumbered-interface-support.patch

James Li jli at cumulusnetworks.com
Sat Oct 26 16:39:57 BST 2013


On 10/26/13 5:58 AM, Joakim Tjernlund wrote:
> Alexis Rosen <quagga-users at alexis.users.panix.com> wrote on 2013/10/25
> 21:43:46:
>> I have no dog in this fight (er, disagreement?) but ISTM that Jocke's
> patches have been in production for years on many many devices. IIRC, his
> use case is not the same as James', but still, that should be worth a lot,
> from the standpoint of wanting to avoid new instabilities, etc.
>
> yes, several years on may hundreds of devices so I am fairly confident
> that they work.
> Many of them needs polish to fit current quagga but they are a good
> starting point.
>
If your patch is not in master, it's not. So you are basically using a 
private version of Quagga. No work can be done in master so that a 
private version won't be affected?
>> I can easily see how Jocke would be unwilling to invest even another
> second in this, as I've seen him try repeatedly over years to get his
> patches reviewed and merged, only to be ignored because the only
> maintainer was unavailable or to absorbed in higher-priority work. Perhaps
> that might change if there were a clear commitment to move forward, one
> way or the other. (And perhaps not, he'll speak for himself I'm sure.)
>
> Exactly, I will not go over this again just to be ignored. Until I see
> some evidence that
> current maintainers show some interest in what is already in patchwork I
> will not yet again
> polish and resubmit for the nth time. I still think Quagga is half asleep
> and moving forward slowly.

Again, I'm done discussing your patch that is not updated. I don't even 
know which patch you are talking about.
> James, as you will find out, there are few devs willing to review stuff
> and even fewer with commit privs.
ok, thanks for the heads up.
> Your "OSPFv2: Support for Unnumbered interface." has just one line
> describing it yet it
> touches 12 files with over 200 additions.
One line? I have sent a supporting document for those who want to read.
> I am sure that there is more going on in there than adding un, numbered I/F
> support, you are
> mixing plain unnumbered ppp with unnumbered eth I/F and ONLINK.
No, it's about unnumbered support in OSPF and Zebra, absolutely nothing 
else. It supports both serial interfaces and LAN interfaces used as PPP, 
which should be a requirement for any implementation attempts.
> What you did to ospf_zebra_add looks really dodgy and broken, the whole
> patch looks odd to say the least.
You have to be more specific. If you didn't review the diff, please 
don't comment.

Anyways, Joakim, I'm done with this thread. If you review, and have real 
review comments, I will happily discuss with you. But the current 
discussion is not productive and will not be continued from my side.

James
>   Jocke
>   
>> /a
>>
>> On Oct 25, 2013, at 1:20 PM, James Li <jli at cumulusnetworks.com> wrote:
>>> Hi list, please review the diff. We know one developer here will not
> review. Hopefully other developers will review and let's push this feature
> forward. After all, Quagga still does not have unnumbered interface
> support after so many years.
>>> Joakim, I'm afraid I cannot engage in another long discussion with you
> about your patch years ago. If you are up to it, you are welcome to bring
> an up-to-date version of your patch against the latest master, compile it,
> test it out, and send it out for review as a competing patch.
>>> If it works and is technically sound, I will be happy to support your
> patch and credit all this work to you. What I want is for unnumbered
> interface to be supported in Quagga.
>>> Before you do that, I will refrain from any further discussion about
> your patch.
>>> thanks,
>>> James
>>>
>>> On 10/25/13 12:27 AM, Joakim Tjernlund wrote:
>>>> James Li <jli at cumulusnetworks.com> wrote on 2013/10/25 01:08:01:
>>>>
>>>>> From: James Li <jli at cumulusnetworks.com>
>>>>> To: Joakim Tjernlund <joakim.tjernlund at transmode.se>,
>>>>> Cc: ayabaner at gmail.com, Dinesh G Dutt <ddutt at cumulusnetworks.com>,
>>>> quagga-dev at lists.quagga.net, sfeldma at cumulusnetworks.com,
>>>> shm at cumulusnetworks.com
>>>>> Date: 2013/10/25 01:08
>>>>> Subject: Re: [quagga-dev 10830] [PATCH 3/4] ospfd:
>>>> ospfd-unnumbered-interface-support.patch
>>>>> On 10/24/13 3:46 PM, Joakim Tjernlund wrote:
>>>>>> James Li <jli at cumulusnetworks.com> wrote on 2013/10/24 23:31:48:
>>>>>>> Hi Jocke, I will send some design notes via another thread, and
> leave
>>>>>>> this one for code review.
>>>>>> Sorry but I will not review this big all in one patch, design notes
> or
>>>>>> not.
>>>>>> The little I saw reminded me of my patches I sent years ago(still
> in
>>>>>> patchwork).
>>>>>> You might want to take a look at these.
>>>>>>
>>>>> This patch is centered around unnumbered interface support, and
> every
>>>>> line of change is an integral part of it. What can be pulled as
> separate
>>>>> patch has been, i.e. the ONLINK and new CLI command "ip ospf area
> ..."
>>>>> are 2 separate patches.
>>>> No, you can and should break it down into smaller logical pieces
> which
>>>> introduces the functionality stepwise. No patch can break the build
>>>> either.
>>>>
>>>>> I don't know what's the history about your patch. Is it accepted and
>>>>> merged into master? If not, and if it's a few years old, then my
>>>>> understanding is it has expired. List masters please correct me if
> this
>>>>> (w.r.t expiration) is not the right understanding.
>>>> I have lots of patches in patchwork, several are still relevant.
>>>> Especially those that fix some bug. They may not apply anymore but
> the
>>>> problem is still there.
>>>> For a long time I tried really hard to get them applied but as Q was
>>>> unmaintained they got dropped on the floor. Now I just point out that
>>>> they are there for anyone who cares to look. Patchwork is basically a
>>>> goldmine
>>>> with stuff but you need to go through it carefully and ask questions.





More information about the Quagga-dev mailing list