Created attachment 161853 [details] [review]
a fix

Seems ok the patch.

Created attachment 161914 [details] [review]
there was a mistake in the other patch + I have added string normalization

Review of attachment 161914 [details] [review]:

After some discussion in IRC, we believe that the history items shouldn't be modified, apart from these two things:
 * removal of leading and trailing whitespaces
 * normalization

Could you then remove the deletion of leading and trailing non-alphanumeric and non-underscore chars? The rules to know if a given word will give results or not are in the parser, and those rules shouldn't be duplicated anywhere, not even for the history items.


@@ -86,6 +86,8 @@ class TrackerSearchEntry  : ComboBoxEntry implements Gtk.Activatable
-             /* remove leading non-alphanumeric chars */
-             while(!txt[0].isalnum())
-                if txt[0] == '_'
-                    break
-                 txt = txt.slice(1, len--)
-                 if len < 3
-                     return false
@@ -93,6 +95,8 @@ class TrackerSearchEntry  : ComboBoxEntry implements Gtk.Activatable
-             /* remove trailing non-alphanumeric chars */
-             if(!txt[len - 1].isalnum())
-                 while(!txt[len - 2].isalnum())
-                    if txt[len - 2] == '_'
-                        break
-                     txt = txt.slice(0, len - 2)
-                     len--
-                     if len < 3
-                        return false

This bug is now obsolete. See bug 619826.

This problem has been fixed in the development version. The fix will be available in the next major software release. Thank you for your bug report.