Re: [PATCH] D65994: Extended FPOptions with new attributes

2020-03-11 Thread Sam McCall via cfe-commits
On Wed, Mar 11, 2020 at 4:09 PM Serge Pavlov wrote: >In this environment, 5 bits are quite a lot for FP flexibility, and I >> think the complexity of reducing this is small and isolated (rounding + >> exceptions together fit in 4 bits) > > > Rounding (5 standard variants) and exception (3 var

Re: [PATCH] D65994: Extended FPOptions with new attributes

2020-03-11 Thread Serge Pavlov via cfe-commits
> >In this environment, 5 bits are quite a lot for FP flexibility, and I > think the complexity of reducing this is small and isolated (rounding + > exceptions together fit in 4 bits) Rounding (5 standard variants) and exception (3 variants) already do not fit 4 bits. And there is also precis

Re: [PATCH] D65994: Extended FPOptions with new attributes

2020-03-11 Thread Sam McCall via cfe-commits
On Wed, Mar 11, 2020 at 12:42 PM Serge Pavlov wrote: > It is a matter of taste, but for me it looks strange to develop complex > and error-prone system to save 7 bits at expense of readability and > maintainability. My observations show that clang AST consumes much less > memory than llvm objects

Re: [PATCH] D65994: Extended FPOptions with new attributes

2020-03-11 Thread Serge Pavlov via cfe-commits
It is a matter of taste, but for me it looks strange to develop complex and error-prone system to save 7 bits at expense of readability and maintainability. My observations show that clang AST consumes much less memory than llvm objects. FPOption could be shared between user using something like s