[quagga-dev 8823] Re: RE-0.99.17.3

Denis Ovsienko infrastation at yandex.ru
Fri Sep 9 14:55:22 BST 2011


09.09.2011, 17:18, "Greg Troxel" <gdt at ir.bbn.com>:
> Can you explain how this corresponds to the git sources, in terms of
> branches and tags?

Yes, sure. It is "quagga-RE" git repository, tag "RE-0.99.17.3":

http://code.quagga.net/?p=quagga-RE.git;a=shortlog;h=refs/heads/RE-testing-0.99

Version 0.99.17.1 exists, but was never released as a tar.gz. 0.99.17.2 turned out to have a regression and was immediately superseded by 0.99.17.3.

Let me take another chance to explain once again, what QRE is and what it is not. I don't have a goal in itself to establish and maintain a fork. And there is no goal in itself to avoid forks at any cost. This work is in the first place about defect management. It is slow, it is demanding, but it's each one's decision if they need the results or not. If in year 2011 it takes a separate git repo with two branches in it to deliver these results, let be it so. If you want to help testing, just join.

I appreciate the pre-0.99.18 work done by Paul WRT ospfd and hope to confirm eventually, that it works for me as good as it worked for him. Efforts _properly_ joined are better, than split. But in my environment one must have a rollback release each time going to upgrade.

Imagine it happens so, that all current commits from master are picked into QRE one by one, tested, each confirmed to be safe (as they are or with fixes applied), then yes, "RE-testing-0.99" is diff-wise a pure derivative from "master", although commits go in a different order in each. In this case it is possible to git-rebase one against the other and make it look nice. But it's critical to make it _work_ nice first, which dictates the order of all other things.

I hope it explains at least something.

-- 
    Denis Ovsienko



More information about the Quagga-dev mailing list