Drastically reduce the transitive dependencies of jsregexp.h, making it (almost)

src/mips/regexp-macro-assembler-mips.h

Index: src/mips/regexp-macro-assembler-mips.h
===================================================================
--- src/mips/regexp-macro-assembler-mips.h	(revision 8581)
+++ src/mips/regexp-macro-assembler-mips.h	(working copy)
@@ -29,6 +29,12 @@
 #ifndef V8_MIPS_REGEXP_MACRO_ASSEMBLER_MIPS_H_
 #define V8_MIPS_REGEXP_MACRO_ASSEMBLER_MIPS_H_
 
+#include "mips/assembler-mips.h"
+#include "mips/assembler-mips-inl.h"
+#include "macro-assembler.h"

[danno, 2011/07/11 08:57:56]

Please include in alpabetical order:

http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Names_and_Orde...

[Sven Panne, 2011/07/11 12:22:48]

IMHO this is a rather useless rule: If you have only a small handful of
#includes, nobody seriously cares about their order, because you can see
everything at first glance. If you have tons of #includes, you have very
different and much more serious problems (coupling/cohesion), where the order is
one of your least concerns. :-)

Anyway, back to the code at hand: The vast majority of our v8 headers fail
miserably when it comes to being self-contained, so we currently *do* have lots
of places where we have to rely on the order.

We really have to clean this up heavily, but this can only be done in very small
steps, because tiny changes ripple through tons of source files at the moment.

+#include "code.h"
+#include "mips/macro-assembler-mips.h"
+
 namespace v8 {
 namespace internal {
 
@@ -249,4 +255,3 @@
 }}  // namespace v8::internal
 
 #endif  // V8_MIPS_REGEXP_MACRO_ASSEMBLER_MIPS_H_
-