Created attachment 187897 [details] [review]
messageTray: improve bad-markup handling

_fixMarkup() was supposed to be ensuring that the markup we passed to
clutter was correct, but it was validating the syntax incorrectly, and
wasn't checking that the markup was valid (or even well-formed). This
is bad because if you pass bad pango markup to
clutter_text_set_markup(), it will g_warn and drop the string on the
floor.

Fix by fixing up the regexps, and then calling Pango.parse_markup() on
the result, and just xml-escaping everything if parse_markup() fails.

Review of attachment 187897 [details] [review]:

Urgh.  Can we get some example test cases for this?  We don't really have a good way to run js unit tests right now I guess, but a few commented examples of what _fixMarkup is supposed to take as input and return as output would be *really* helpful for review and future bugfixes.

Created attachment 188108 [details] [review]
tests: fix up typelib include paths

The js modules have so many imports back and forth that it's pretty
much guaranteed that if you import even one of them, you'll end up
importing all of them, including ui.status.bluetooth and
ui.status.network. So fix up the typelib include paths the same way
gnome-shell-jhbuild does, so we can find everything.

Created attachment 188109 [details] [review]
tests: add a test for MessageTray._fixMarkup

Review of attachment 188108 [details] [review]:

::: tests/run-test.sh.in
@@ -31,3 @@
 srcdir=`cd $srcdir && pwd`
 
-GI_TYPELIB_PATH="@MUTTER_TYPELIB_DIR@:$builddir/../src"

UPowerGlib was removed from gnome-shell.modules in favour to system one.
/usr/{lib64,lib}/girepository-1.0 should be in GI_TYPELIB_PATH.

Review of attachment 188109 [details] [review]:

looks good

::: tests/unit/markup.js
@@ +86,3 @@
+
+// stray '&'; this is really a bug, but it's easier to do it this way
+assertConverts('<b>this</b> & <i>that</i>', '<b>this</b> &amp; <i>that</i>'); 

trailing whitespace

Created attachment 192667 [details] [review]
tests: fix up typelib include paths

The UPowerGLib change shouldn't affect anything; g-i looks in the
system dirs anyway. But the code was failing to use the existing value
of GI_TYPE_LIB_PATH, so fix that.

Created attachment 192668 [details] [review]
tests: add a test for MessageTray._fixMarkup

fix trailing whitespace

Attachment 187897 [details] pushed as 7542e68 - messageTray: improve bad-markup handling
Attachment 192667 [details] pushed as 0907f03 - tests: fix up typelib include paths
Attachment 192668 [details] pushed as f03793b - tests: add a test for MessageTray._fixMarkup