|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [Xen-devel] add_maintainers.pl issues / "canonical" subject prefixes for CI v2 (was Re: [PATCH] create-diff-object: more precisely identify .rodata sections)
+Ross
+ Doug
On 19/09/2019, 10:30, "Julien Grall" <julien.grall@xxxxxxx> wrote:
Hi Lars,
On 18/09/2019 13:14, Lars Kurth wrote:
>
>
> On 18/09/2019, 12:15, "Julien Grall" <julien.grall@xxxxxxx> wrote:
>
> Hi Ian,
>
> On 18/09/2019 11:41, Ian Jackson wrote:
> > Julien Grall writes ("Re: [PATCH] create-diff-object: more
precisely identify .rodata sections"):
> >> On 18/09/2019 10:52, Wieczorkiewicz, Pawel wrote:
> >>> $ scripts/./add_maintainers.pl -d ~/git/livepatch-build-tools
> >>
> >> '-d' only tells you where the patches files are. The script will
look up for the
> >> MAINTAINERS file in the current directory.
> >
> > Hmmm. I wonder if we could detect this situation somehow. This
will
> > be a common user error I think.
> I think it would be possible for patch modifying file. We could
check whether
> the file modified exist in the repo. Though, I am not sure how
difficult it
> would be to implement.
>
> That might be doable, but won't be easy as I will essentially need to
parse the patch
> And it won't be reliable.
>
> The only workable way of doing this may be to have a strong convention
> that requires to use the [REPONAME PATCH] via --subject-prefix when
generating the
> patch and for add_maintainers.pl to verify this somehow based on the
current
> directory and the patches.
>
> We already have strong conventions in some cases, e.g. for OSSTEST we
always use
> [OSSTEST PATCH]. This would potentially be helpful for the CI loop plans
aso.
>
> Assuming there is a git config setting for --subject-prefix then this
could be made
> to work. I could add a section under [1] to document the convention with
the
> appropriate git command. We could include a script (e.g.
xen.git:scrips/git-setup)
> which does this based on the repo name automatically.
I saw a conversation on IRC about this. But it is not clear if there was
any
conclusion?
Ian was going to write down what we discussed
My only slight concern about tagging by default is the length of the
subject,
when directly receiving from xen-devel (i.e. not CCed), the subject is
already
[xen-devel][PATCH XX/XX]. I am assuming the tag will not be dropped so it
will
appear on the mailing list. For repo such as LIVEPATCH-BUILD, this would
end up
to lengthy prefix.
That is true: I think we do need to have a formal discussion with vote about
this at some point.
And we can certainly shorten some of those "canonical" subject prefixes.
But having gone back to the early patches posted for LIVEPATCH-BUILD by Konrad
and Ross, they were either
[LIVEPATCH-BUILD-TOOLS] or [LIVEPATCH-BUILD-TOOLS PATCH]
My view at this stage, is that I am merely documenting existing conventions
The exception is "XEN PATCH"
For people who are just starting to read this conversation, see
*
https://wiki.xenproject.org/wiki/Submitting_Xen_Project_Patches#Settings_that_help_save_you_time
*
https://wiki.xenproject.org/wiki/Submitting_Xen_Project_Patches#Subject_Prefix
which I added yesterday
We do need to solve this somehow for two reasons:
1. Fool proofing add_maintainers.pl tool [although this is relatively minor]
2. More importantly we need to support the CI v2 capability (aka triggering
tests for patches posted to xen-devel@)
On 1: Ian had an idea, which I think I understood correctly, but am not 100%
sure. So waiting for him to write it down
On 2: We need a way to identify which tree a patch or patch series is for, such
that any CI infra can perform the right action, in light of
* QEMU, LINUX, ... patches are posted to xen-devel@ today
* Patches for multiple xen-repos are posted to xen-devel@ today
But maybe we should move this into a new thread. In the meantime I changed the
subject line
Lars
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel
|
![]() |
Lists.xenproject.org is hosted with RackSpace, monitoring our |