[WebIDL] Do a pass of cleanup in the bindings generator
Created attachment 311228 [details] Patch
Comment on attachment 311228 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=311228&action=review > Source/WebCore/bindings/js/JSReadableStreamSourceCustom.cpp:42 > + // FIXME: Why is it ok to ASSERT the argument count here? I don't think it is OK :S
Committed r217433: <http://trac.webkit.org/changeset/217433>
(In reply to Chris Dumez from comment #2) > Comment on attachment 311228 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=311228&action=review > > > Source/WebCore/bindings/js/JSReadableStreamSourceCustom.cpp:42 > > + // FIXME: Why is it ok to ASSERT the argument count here? > > I don't think it is OK :S Youenn says it is "safe" because this API is only called from internal JS, not exposed to the Web. Either way, I still think this is very unnecessarily fragile and we should do an arg count check. Right now, it is calling uncheckedArgument() and would likely crash.
JSReadableStreamSourceCustom start/pull is only to be called by ReadableStream implementation which will always pass a controller. It is safe to do so and all fetch API+body as stream tests would crash if we were to have an issue there. If you want more safety here, removing uncheckedArgument and ensuring we got a controller would be needed.
<rdar://problem/32479796>