[llvm] r251209 - [CodeGen] Get rid of NDEBUG to ensure structure stability.

[Davide Italiano via llvm-commits]

On Sun, Jan 10, 2016 at 5:19 PM, Duncan P. N. Exon Smith
<dexonsmith at apple.com> wrote:
>
>> On 2016-Jan-06, at 15:43, Davide Italiano <davide at freebsd.org> wrote:
>>
>> On Fri, Nov 20, 2015 at 12:00 PM, Duncan P. N. Exon Smith
>> <dexonsmith at apple.com> wrote:
>>>> On 2015-Oct-24, at 15:09, Davide Italiano via llvm-commits <llvm-commits at lists.llvm.org> wrote:
>>>>
>>>> Author: davide
>>>> Date: Sat Oct 24 17:09:54 2015
>>>> New Revision: 251209
>>>>
>>>> URL: http://llvm.org/viewvc/llvm-project?rev=251209&view=rev
>>>> Log:
>>>> [CodeGen] Get rid of NDEBUG to ensure structure stability.
>>>>
>>>> I think it's fine to keep this fields around in terms of overhead,
>>>> I wasn't able to measure any substantial regression while running the
>>>> test suite, but, in case this causes some regression I'm ready to revert
>>>> and work on an alternative solution.
>>>> This was tested building with clang/gcc both in Debug and Release mode
>>>> and passes the test-suite.
>>>>
>>>
>>> I wonder if adding to `SDNode` could be a problem for memory-constrained
>>> backends that tend to only have a single function?
>>>
>>> This could use LLVM_ENABLE_ABI_BREAKING_CHECKS (see r233310) instead
>>> of NDEBUG, if so.
>>>
>>
>> Duncan, I got to this again now (hopefully not too late). Is it fine
>> to leave it as is? Or you want me to do the change? Resistor, as owner
>> of this subsystem, what's your thought?
>
> Seems better to save the memory with LLVM_ENABLE_ABI_BREAKING_CHECKS
> since it's not too hard.  Any thoughts Owen?

OK I'll post the patch once phabricator will come back online.

Thanks!

-- 
Davide

"There are no solved problems; there are only problems that are more
or less solved" -- Henri Poincare