This is the mail archive of the
mailing list for the GCC project.
Re: [lto] PATCH: Add support for referencing types from function bodies
Mark Mitchell wrote:
> Kenneth Zadeck wrote:
>> This is exactly what I did not want to do. Because now the output
>> gimple will be different from the input gimple and it will make it more
>> difficult to verify that I am doing the lto reading and writing
> I don't think there's a perfect choice here. You reasonably want the
> GIMPLE to be identical on read/write, but I, also reasonably, don't
> want to write code we don't need in the long term.
> You/I/we could instead teach the gimplifier on the LTO branch to make
> the transformation from CONST_DECL to its initial value. That would
> avoid transforming the GIMPLE as you write it.
I think that if you want to teach the gimplifier to do this, that is a
great way to get rid of them for any downstream part of the compiler. I
have never touched the gimplifier and really do not want to make this
yet another learning experience, dwarf and gimple have been enough for