Suppress a GTK warning and file a bug for it.

chrome/app/chrome_dll_main.cc

Index: chrome/app/chrome_dll_main.cc
===================================================================
--- chrome/app/chrome_dll_main.cc	(revision 14624)
+++ chrome/app/chrome_dll_main.cc	(working copy)
@@ -159,12 +159,19 @@
   if (!message)
     message = "<no message>";
 
-  // http://code.google.com/p/chromium/issues/detail?id=9643
+  // http://crbug.com/9643

[Evan Martin, 2009/04/28 21:13:16]

I think the other (original) URL style is better here and below.  The short one
is just for typing.

[Evan Stade, 2009/04/28 21:23:51]

I don't see why. The short form seems easier to read. Is crbug.com likely to go
away?

A grep of src/chrome/ shows 62 instances of the short form and 44 of the long
form.

   // Until we have a real 64-bit build or all of these 32-bit package issues
   // are sorted out, don't fatal on ELF 32/64-bit mismatch warnings.
   if (strstr(message, "Loading IM context type") ||
       strstr(message, "wrong ELF class: ELFCLASS64")) {
     LOG(ERROR) << "Bug 9643: " << log_domain << ": " << message;
+  // http://crbug.com/11133
+  // gtk_check_version() will return NULL if the current version is >= the

[Evan Martin, 2009/04/28 21:39:51]

be sure to remove this comment once you use the macro.

I think it would be less confusing to have the bug link inside the else branch,
but that's your call.

+  // passed version.
+  } else if (strstr(message, "gtk_widget_size_allocate(): attempt to "
+                             "allocate widget with width") &&
+             gtk_check_version(2, 16, 1)) {

[Evan Martin, 2009/04/28 21:13:16]

Why not GTK_CHECK_VERSION?
"Returns TRUE if the version of the GTK+ header files is the same as or newer
than the passed-in version."

I wonder if this will run into trouble with translated messages... :(

[Evan Stade, 2009/04/28 21:23:51]

Done.

+    LOG(ERROR) << "Bug 11133";
   } else {
     LOG(FATAL) << log_domain << ": " << message;
   }