> Considering the amount of patches you needed to post, the fact we don't have a good handle on this feature, and that the tests weren't actually giving us proper feedback, we should give up on this for 9.0 at this very late stage.
Note that the number of patches isn't an indication of complexity, I just split the change up into granular commits.
That said, I tend to agree that given the lack of testing, it's a risky change to put into RC4.
> Are we currently dealing with a half working situation? We need to turn it off in a consistent manner so it can be brought in for 9.1. Are any changes needed for that? Could they cause more damage than good?
I think it might work in some cases, but, worryingly, it can access the index without acquiring the index read lock.
It should be safe to disable it the way Sergey disabled it before while investigating a performance regression [1]. I can put together a patch tonight after I get home from my day job.
Regards,
Nate
[1]
http://git.eclipse.org/c/cdt/org.eclipse.cdt.git/commit/?id=4ac6657792353e6fd473e2f6c34baaf0538d1261
_______________________________________________
cdt-dev mailing list
cdt-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/cdt-dev
