Bug 353121

Created attachment 237881 [details]
ignore duplicates when adding spelling error objects to a line's spell_errors list

Using a 15375 character long line of multiple words (me pressing asdfgwer and
space at random times) I couldn't determine a performance difference between
patched and unpatched versions with a stopwatch.  Each text line keeps track of
it's own spell_list, so the speed impact of searching through the spell_errors
list is usually nothing since most lines will have < 10 spelling errors, even if
they are computery things like patches with lots of acronyms and non-english words.

Ray; can you take a quick look at the patch too for more assurance that it's
correct?  Does linked list manipulation and I want to be sure it looks OK to a
few people :)

Looks like it should work.  An alternate approach that avoids the linked list
manipulation might be to define a spell_errors_list_merge() function that
consumes two sorted GLists returns a new sorted GList, free of dupes.

Something roughly like (in Python):

    def spell_errors_list_merge(L1, L2):
            L3 = []
            while L1 and L2:
                if L1[0] < L2[0]:          # se_cmp() < 0
                    n = L1.pop(0)          # g_list_remove_link()
                else:
                    n = L2.pop(0)          # g_list_remove_link()
                if not L3 or n != L3[0]:   # se_cmp() != 0
                    L3.insert(0, n)        # g_list_prepend()
            L3.reverse()                   # g_list_reverse()
            L3.extend(L1)                  # g_list_concat()
            L3.extend(L2)                  # g_list_concat()
            return L3

Pushing this upstream with a revised patch.  See [1] for further updates.

[1] http://bugzilla.gnome.org/show_bug.cgi?id=495073