[8u] RFR: cumulative patch required for C1 integration
Andrey Petushkov
andrey.petushkov at gmail.com
Wed Feb 17 16:08:37 UTC 2016
Hi Andrew,
Thank you for your response. Yes, sure, we understand the changes are rather massive. Actually this patch is a reworked merge of 158 c1-related changesets we have made to the moment. So please try to walk in our shoes - we have a different code baseline (we have taken up from latest stable release at that moment, which was 8u60) and almost every of these 158 patches touch both shared and c1-specific files. So in order to prepare what you’re asking for we have to rebeaseline all of those 158, but then you’ll end up in 158 review requests on the alias. It’s definitely much easier to review each single one of them but really do you want to take all of them one by one? Then logically each of these 158 will need a separate review for the c1-specific files, however the latter is nonsense, since it’s not a bugfix or improvement of some existing code but rather a major rework of (aarch64 c1 implementation).
Sometimes in course of our development we were going back and forth around the same thing, changing the same lines of code in several commits. Of course it does not make sense to review these commits one by one. At the same time in order to prepare a self-contained complete changeset for particular problem/feature it would require to carefully handcraft the patch from the several of those 158. Which tend to take amount of work comparable to fixing the bugs/implementing the features themselves!
So provided all of that how would you suggest to approach the task of publishing our c1 related works into the open?
Thanks,
Andrey
> On 17 Feb 2016, at 18:24, Andrew Haley <aph at redhat.com> wrote:
>
> On 02/17/2016 03:10 PM, Sergey Nazarkin wrote:
>> Below are rough descriptions of what has been introduced in the changeset:
>
> Umm, this is great work! Thanks.
>
> However, putting all this in a single changeset without public
> discussion is not such great work. We have no way to know what he
> problems were. How then shall we review the changes in a meaningful
> way?
>
> I don't want to appear ungrateful, but you've just thrown a 9,000 line
> patch over the wall.
>
> Andrew.
More information about the aarch32-port-dev
mailing list