New avatar button: Consolidate text elision between Mac and Win/Linux

ui/gfx/text_elider.h

Index: ui/gfx/text_elider.h
diff --git a/ui/gfx/text_elider.h b/ui/gfx/text_elider.h
index a416fda779eb37e42ff32d2dbadb625ffcc62a3c..cd898c1853d965ba6ed2d3d7be24d42e60318324 100644
--- a/ui/gfx/text_elider.h
+++ b/ui/gfx/text_elider.h
@@ -158,7 +158,8 @@ GFX_EXPORT int ElideRectangleText(const base::string16& text,
 // The supplied string is returned if the string has length characters or
 // less.
 GFX_EXPORT base::string16 TruncateString(const base::string16& string,
-                                         size_t length);
+                                         size_t length,
+                                         bool only_at_word_breaks);

[msw, 2014/07/17 17:18:35]

nit: maybe use gfx::BreakType or base::i18n::BreakIterator::BreakType?

[Marc Treib, 2014/07/18 10:05:09]

Hm, base::i18n::BreakIterator::BreakType has lots of cases that aren't handled
separately here, and it might be confusing that the behavior here doesn't quite
match that of a corresponding BreakIterator.
gfx::BreakType has WORD_BREAK and LINE_BREAK which would probably mean the same
thing here? But there's a TODO to merge it with
base::i18n::BreakIterator::BreakType, so.. hm.
I agree that an enum would be better here than a bool, but neither of these seem
to really fit, and I wouldn't like to introduce yet another one.

[msw, 2014/07/18 19:21:48]

gfx::BreakType seems appropriate for now. When my TODO is addressed, it can be
replaced with a [likely leaner] base::i18n::BreakIterator::BreakType. It already
supported WORD_BREAK and now supports CHARACTER_BREAK, and could later support
LINE_BREAK (start by finding the first newline before the argument length). If
we are adding an argument to support some addition breaking type, it should be
an enum, not a bool.

[Marc Treib, 2014/07/21 09:03:00]

Done.

 
 }  // namespace gfx