[llvm] r246367 - [Triple] Stop abusing a class to have only static methods and just use

Renato Golin via llvm-commits llvm-commits at lists.llvm.org
Sun Aug 30 03:47:14 PDT 2015


On 30 August 2015 at 11:20, Chandler Carruth <chandlerc at gmail.com> wrote:
> This is a very simple change, and no one seemed actively modifying it, so I
> don't think there is a problem here...

As I said on the other thread, small changes, slow paced.


> Even if there are some future plans, we shouldn't design the code in a way
> that makes no sense today because we think (without evidence) that it will
> make more sense in the future. YAGNI, or insert your software design
> aphorism here.

As I said before, this has been discussed openly on the list. It's not
because you haven't participated that there is no evidence.



> This code, today, is doing procedural text parsing, and as such should just
> be using functions in a name space.

Maybe.


>If and when some future design is
> actually proposed, I think that would be the time to change it.

The future design was proposed, and is being implemented as part of
this is the TargetTuple design, which the TargetParser is waiting on,
but Eric is blocking the review for the last months, so we couldn't
progress as fast as we wanted.



> I don't think that is the correct approach. This patch in particular I feel
> *extremely* confident is a strictly positive step forward.

I don't. What do we do now?


> Now if you want to remove *all* of this code until that better state is
> identified, that makes more sense but I think that ship has sailed. That
> would be perceived as an acceptable regression. The only path forward I see
> here is to make the code that we have today be reasonable, and ask you and
> others to try to come up with a better design long term. I think these
> patches are reasonable steps in the first goal.

Shouldn't we have discussed this *before* the patches?

Is it ok for me to start changing other people's areas at will just
because "I'm extremely confident" that I'm right?

Wouldn't you have the right to revert my patches if that would break
your line of work?

I'm feeling a big heavy hand here, and I'm not liking it much. Or is
this the new state of LLVM and we (minor contributors) should just
accept and move along?

cheers,
-renato


More information about the llvm-commits mailing list