write w/o approval policy (Re: [PATCH] clarify comments for implicit_p flag for built-ins)

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

write w/o approval policy (Re: [PATCH] clarify comments for implicit_p flag for built-ins)

Martin Sebor-2
On 11/28/18 6:35 AM, Richard Biener wrote:
> On Tue, Nov 27, 2018 at 3:52 AM Martin Sebor <[hidden email]> wrote:
>>
>> Ping: https://gcc.gnu.org/ml/gcc-patches/2018-11/msg01759.html
>>
>> If there are no objections or suggestions for tweaks I'll commit
>> this updated comment this week.
>
> Please do not commit such changes w/o approval.

Since you're the second maintainer to ask me that in response
to a patch to update comments I'd like to get some clarity here.

I have been assuming that the GCC Write access policy (quoted
below) lets those of us with write-after-approval make a judgment
call as to when a change is sufficiently safe to commit:

   Obvious fixes can be committed without prior approval.  Just
   check in the fix and copy it to gcc-patches. A good test to
   determine whether a fix is obvious: "will the person who
   objects to my work the most be able to find a fault with my
   fix?"  If the fix is later found to be faulty, it can always
   be rolled back. We don't want to get overly restrictive about
   checkin policies.

   (https://www.gnu.org/software/gcc/svnwrite.html#policies)

If we are not at liberty to make this judgment call in even
the most innocuous cases like comments, when does this policy
actually apply?  (It should be updated to make it clear.)

Thanks
Martin
Reply | Threaded
Open this post in threaded view
|

Re: write w/o approval policy (Re: [PATCH] clarify comments for implicit_p flag for built-ins)

Jeff Law
On 11/28/18 11:39 AM, Martin Sebor wrote:

> On 11/28/18 6:35 AM, Richard Biener wrote:
>> On Tue, Nov 27, 2018 at 3:52 AM Martin Sebor <[hidden email]> wrote:
>>>
>>> Ping: https://gcc.gnu.org/ml/gcc-patches/2018-11/msg01759.html
>>>
>>> If there are no objections or suggestions for tweaks I'll commit
>>> this updated comment this week.
>>
>> Please do not commit such changes w/o approval.
>
> Since you're the second maintainer to ask me that in response
> to a patch to update comments I'd like to get some clarity here.
>
> I have been assuming that the GCC Write access policy (quoted
> below) lets those of us with write-after-approval make a judgment
> call as to when a change is sufficiently safe to commit:
>
>   Obvious fixes can be committed without prior approval.  Just
>   check in the fix and copy it to gcc-patches. A good test to
>   determine whether a fix is obvious: "will the person who
>   objects to my work the most be able to find a fault with my
>   fix?"  If the fix is later found to be faulty, it can always
>   be rolled back. We don't want to get overly restrictive about
>   checkin policies.
>
>   (https://www.gnu.org/software/gcc/svnwrite.html#policies)
>
> If we are not at liberty to make this judgment call in even
> the most innocuous cases like comments, when does this policy
> actually apply?  (It should be updated to make it clear.)
The thing is I looked at the patch and it was far from obvious what was
going on.  Thus I put it in my queue of things to dig deeper into.  I
haven't done that digging yet.

Comments are actually important.  They often describe what the code is
supposed to do, rationale, historical context, etc.  Just because we're
changing a comment doesn't mean it's inherently trivial/obvious.

I'm generally supportive of lessening friction for developers and I
welcome proposals to do that.

Jeff