Re: 回复: 回复: [edk2-devel] please let's disable the ECC plugin in CI, in its current form

[Sean]

I have been thinking about the idea of a CI plugin type that is provided 
a file list and only operates on the supplied file list.

Then during CI execution if a file list was provided these plugins would 
run and evaluate on that file list.  In azure pipeline we already get 
the diff between target branch and PR.  We could then write that to file 
and pass that file into the stuart_ci_build process.

Anyway if you have other ideas i'll leave it to you.  Please let me know 
if you are interested, want to discuss more, or go in this direction.

Thanks
Sean


On 12/9/2020 5:34 PM, gaoliming wrote:
> Sean:
>   I will give the proposal to fix this issue https://bugzilla.tianocore.org/show_bug.cgi?id=2986. In fact, we can enhance ECC to skip the folder or files.
> 
> Thanks
> Liming
>> -----邮件原件-----
>> 发件人: Sean Brogan <spbrogan at outlook.com>
>> 发送时间: 2020年12月9日 2:37
>> 收件人: devel at edk2.groups.io; gaoliming at byosoft.com.cn;
>> lersek at redhat.com; 'Michael Kinney' <michael.d.kinney at intel.com>; 'Sean
>> Brogan' <sean.brogan at microsoft.com>; 'Ard Biesheuvel (ARM address)'
>> <ard.biesheuvel at arm.com>; 'Leif Lindholm (Nuvia address)'
>> <leif at nuviainc.com>; 'Andrew Fish' <afish at apple.com>
>> 主题: Re: 回复: [edk2-devel] please let's disable the ECC plugin in CI, in its
>> current form
>>
>> Liming
>>
>> I agree it would be great to get to a place where a package maintainer
>> could turn this on but because of this bug
>> https://bugzilla.tianocore.org/show_bug.cgi?id=2986 I think this needs
>> to be completely disabled until the above is resolved.
>>
>> Thanks
>> Sean
>>
>>
>> On 12/8/2020 7:38 AM, gaoliming wrote:
>>> Laszlo:
>>>     ECC tool is not perfect. It is still helpful. I am OK to disable ECC plugin in
>> CI. But, I request to expose the option to enable it per package. If the
>> package maintainer thinks ECC is valuable, he can enable ECC plugin in one
>> package. As MdePkg maintainer, I can help to resolve ECC issue in MdePkg
>> patch. So, I would keep ECC plugin in MdePkg.
>>>
>>> Thanks
>>> Liming
>>>> -----邮件原件-----
>>>> 发件人: bounce+27952+68414+4905953+8761045 at groups.io
>>>> <bounce+27952+68414+4905953+8761045 at groups.io> 代表 Laszlo
>> Ersek
>>>> 发送时间: 2020年12月8日 10:05
>>>> 收件人: Michael Kinney <michael.d.kinney at intel.com>; Sean Brogan
>>>> <sean.brogan at microsoft.com>; Ard Biesheuvel (ARM address)
>>>> <ard.biesheuvel at arm.com>; Leif Lindholm (Nuvia address)
>>>> <leif at nuviainc.com>; Andrew Fish <afish at apple.com>
>>>> 抄送: edk2-devel-groups-io <devel at edk2.groups.io>
>>>> 主题: [edk2-devel] please let's disable the ECC plugin in CI, in its current
>> form
>>>>
>>>> Hi All,
>>>>
>>>> in my opinion, the ECC plugin in CI has not been productive or helpful.
>>>> The errors it reports are not convincing, and exceptions are difficult
>>>> to add, or even express. I request that we disable ECC globally for
>>>> edk2, or at least make it controllable through a github PR flag (not
>>>> through CI metafile changes in edk2).
>>>>
>>>> (Metafile changes could be an acceptable way of controlling ECC if (a)
>>>> they didn't have to go through review, (b) if technically such changes
>>>> would take effect in CI even if they were included in the patch series
>>>> subject to the CI run. Then a maintainer could evaluate and
>>>> *immediately* suppress such ECC issues by adding the exceptions as a
>>>> prepended patch, and force-pushing the updated branch to the same open
>>>> PR. But this would remain much inferior to simply disabling ECC, or
>>>> controlling it through a PR label.)
>>>>
>>>> Thanks
>>>> Laszlo
>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
> 
> 
> 
> 
> 
> 
> 


-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#68630): https://edk2.groups.io/g/devel/message/68630
Mute This Topic: https://groups.io/mt/78845172/1813853
Group Owner: devel+owner at edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [edk2-devel-archive at redhat.com]
-=-=-=-=-=-=-=-=-=-=-=-