I am on vacation so I won't be able to review your patch in detail, but from your description it sounds less appealing to me than the WKTR approach. It seems like bad layering to me to define the IDL interface in WebCore for something actually implemented completely outside of WebCore. It also seems better to use the public interface to the JS engine rather than the internal interface, even though this requires additional codegen back ends (one of which is already written). But, I may be missing advantages to your approach, since I only gave it a quick glance.
- Maciej On Jun 3, 2012, at 5:17 PM, Ryosuke Niwa <rn...@webkit.org> wrote: > On Sun, Jun 3, 2012 at 1:45 PM, Maciej Stachowiak <m...@apple.com> wrote: > Already done for the WebKit2 WebKitTestRunner. It should be straightforward > to adapt its IDL compiler. > > I've considered adapting WebKitTestRunner's code generator, but it appeared > quite tricky because ports don't share much code in DumpRenderTree unlike > WebKitTestRunner so we end up having to significantly modify each port's > build configuration files for DumpRenderTree. Also, WebKitTestRunner's code > generator only supports JSC at the moment, and supporting V8 there will > require a significant work and might clutter the codebase. > > Given that, the approach I'm taking in > https://bugs.webkit.org/show_bug.cgi?id=88183 at the moment is re-use > Internals object's infrastructure. So I'm defining > LayoutTestControllerBase.idl (interfaceName=LayoutTestController) and > LayoutTestControllerBase.h/cpp in WebCore, and letting LayoutTestController > in DumpRenderTree override it. > > One big advantage of this approach is that code generator in WebCore already > supportsand V8, so we'll be able to share more code between JSC & V8 > implementations of LayoutTestController. In fact, my current plan is to > replace all usage of JSStringRef, etc... (and their V8 equivalents) by > std::string so that they can be identical between two implementation > Of course, one big disadvantage is that we'll be introducing a separate IDL > file for LayoutTestController. However, it appears that we can use the same > object injection mechanism for WebKitTestRunner as well. So I'm hopeful that > we can ultimately share IDLs between DumpRenderTree and WebKitTestRunner. > > What do you think of this approach? > > - Ryosuke >
_______________________________________________ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo.cgi/webkit-dev
