CMake on Mac should not affect the decision here.  I added that as a hack in as part of an experimental project that isn’t 
being used by anyone.  If we decide to resume CMake on Mac development and that 
has moved, we will find a better way to solve the same build problem.

> On Dec 8, 2018, at 3:22 PM, Darin Adler <> wrote:
> OK, here’s my answer after thinking on it a bit:
> Best would be to eliminate “config.h”: Change “config.h” into an empty file 
> first, then remove all “config.h” includes, and then remove the file. But to 
> do that, we need to make sure every build system for WebKit supports prefix 
> headers. I don’t know how close to that we are. Maybe close? How can we 
> quickly find out?
> Lacking that, I think we can and should change “config.h” so it’s just an 
> include of “WebCorePrefix.h”, or the other way around. I think it would be 
> valuable to keep the feature where we try to catch cases where we forget to 
> include “config.h”, on the platforms that use a prefix header, for the 
> benefit of the platforms that do not. That might mean small complexity 
> remains and one file won’t literally just be a trivial include of the other.
> I suppose it’s also important to verify that there is no benefit to 
> precompiling less than all of what “config.h” includes.
> — Darin
> PS: I don’t think we know that there is only one configuration that needs 
> “config.h”. That second code snippet from your original message, Alexey, was 
> only relevant for platforms that are trying to support macOS without prefix 
> headers, and there could be any number of non-macOS cases. (And that include 
> seems like a relatively recent change done by someone who didn’t fully 
> understand the original scheme.)
> _______________________________________________
> webkit-dev mailing list

webkit-dev mailing list

Reply via email to