[MPTCP] Re: MPTCP repository suggestion: 'fixes' and 'fixes-next' branches

[MPTCP] Re: MPTCP repository suggestion: 'fixes' and 'fixes-next' branches

[Florian Westphal]

[-- Attachment #1: Type: text/plain, Size: 805 bytes --]

Mat Martineau <mathew.j.martineau(a)linux.intel.com> wrote:
> 
> Now that we have a reasonable amount of code in net-next (and, soon enough,
> in https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net.git), it would
> help to have a couple of branches that bypass the topgit tree:
> 
> fixes-next: Patches staged for the net-next tree
> fixes: Patches staged for the net tree

Either that or follow the model used by other sub-maintainers, i.e.

add mptcp.git (tracking net.git), and mptcp-next.git (for net-next).

> This would give a location to share pending changes and run CI. Seem like a
> good idea? I thought we discussed this in a meeting at some point but didn't
> see it in the notes.

Yes, good idea.  I do not have a preference of branches vs. completely
distinct trees.

[MPTCP] Re: MPTCP repository suggestion: 'fixes' and 'fixes-next' branches

[Matthieu Baerts]

[-- Attachment #1: Type: text/plain, Size: 2041 bytes --]

Hi Mat, Florian,

On 25/01/2020 00:13, Florian Westphal wrote:
> Mat Martineau <mathew.j.martineau(a)linux.intel.com> wrote:
>>
>> Now that we have a reasonable amount of code in net-next (and, soon enough,
>> in https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net.git), it would
>> help to have a couple of branches that bypass the topgit tree:
>>
>> fixes-next: Patches staged for the net-next tree
>> fixes: Patches staged for the net tree
> 
> Either that or follow the model used by other sub-maintainers, i.e.
> 
> add mptcp.git (tracking net.git), and mptcp-next.git (for net-next).

Maybe easier to work with branches? It's just because "mptcp.git" repo 
on Github currently points to the out-of-tree implementation. But we can 
also find new names.
The MAINTAINERS file has a ref to "mptcp_net-next.git" repo.

If we go with branches, we can have "fixes-net" and "fixes-net-next" to 
avoid confusions.

We can also put new commits at the "beginning" of the export branch if 
we are not in a hurry.
But I guess in general we don't need TopGit for these fixes because we 
want to send the ones for net and net-next ASAP, we don't plan to stack 
fixes for long.

>> This would give a location to share pending changes and run CI. Seem like a
>> good idea? I thought we discussed this in a meeting at some point but didn't
>> see it in the notes.
> 
> Yes, good idea.  I do not have a preference of branches vs. completely
> distinct trees.

Sorry, I don't remember about this discussion but indeed, that would be 
easier for the CI if the fixes are stored in Git. Maybe having a hook to 
start a test automatically when one of these branches are updated by one 
of us? We could also rely public CI like Travis if we don't have to 
modify the git repo. The current CI job needs write access.

Cheers,
Matt
-- 
Matthieu Baerts | R&D Engineer
matthieu.baerts(a)tessares.net
Tessares SA | Hybrid Access Solutions
www.tessares.net
1 Avenue Jean Monnet, 1348 Louvain-la-Neuve, Belgium