1
1
# Generated by Makefile. Do not edit.
3
2012-04-05 Charles Kerr <charles.kerr@canonical.com>
7
2012-04-05 Charles Kerr <charles.kerr@canonical.com>
9
merge lp:~desrt/indicator-appmenu/emit-unregister-signal to properly emit the WindowUnregistered signal when windows go away
11
2012-04-04 Ryan Lortie <desrt@desrt.ca>
13
appmenu registrar: properly emit the "WindowUnregistered" signal when windows go away
15
2012-04-04 Charles Kerr <charles.kerr@canonical.com>
17
merge lp:~desrt/indicator-appmenu/no-search-tokens to check for search with no tokens before searching.
19
The search algorithm assumes that there will be at least one search token, so be sure not to submit any searches with zero tokens. This can happen if the user searches for " ", for example.
21
2012-04-04 Ryan Lortie <desrt@desrt.ca>
23
hud-service: check for search with no tokens before searching
25
The search algorithm assumes that there will be at least one search
26
token, so be sure not to submit any searches with zero tokens. This can
27
happen if the user searches for " ", for example.
29
2012-04-04 Charles Kerr <charles.kerr@canonical.com>
31
merge lp:~charlesk/indicator-appmenu/remove-dead-file to remove a file that should have been deleted in the previous commit but was incorrectly resolved. h/t seb128
33
2012-04-04 Charles Kerr <charles.kerr@canonical.com>
35
remove file that should have been deleted in previous commit but was incorrectly resolved. h/t seb128
37
2012-04-03 Charles Kerr <charles.kerr@canonical.com>
39
merge lp:~allanlesage/indicator-appmenu/TDD to remove a dependency on gcovr, move gcov targets to their own makefile, and clean up .gcno files.
41
2012-03-27 Allan LeSage <allanlesage@gmail.com>
43
Pedantic name change for gcovr xml coverage results.
45
2012-03-27 Allan LeSage <allanlesage@gmail.com>
47
Removed gcovr dependency, cleaned up *.gcno files.
49
2012-04-03 Charles Kerr <charles.kerr@canonical.com>
51
merge lp:~desrt/indicator-appmenu/hud-performance to improve hud-service performance and search result quality by substantially reworking the core distance algorithm and the needle/haystack assignment algorithm.
53
2012-04-03 Ryan Lortie <desrt@desrt.ca>
55
HudWindowSource: fix a couple of leaks
57
- don't leak icon name while changing windows (fixes lp:972354)
59
- don't leak the collector on finalize (which never happens)
61
2012-04-02 Ryan Lortie <desrt@desrt.ca>
63
Adjustments for Ted's review comments.
65
2012-04-02 Ryan Lortie <desrt@desrt.ca>
67
Increase the default swap penalty to 15
69
The current value is too low. It allows "xyze" as a valid match for "File" (3
70
swapped characters at a cost of 10 each gives a total cost of 30, which is the
71
maximum permitted value).
73
It also causes an issue with accents. Consider the case of typing
74
"pref" and encountering the menu items "préférence" and "précédent".
75
Clearly the user wants "préférence". During normalisation, we convert
76
each of these strings to "pre'fe'rence" and "pre'ce'dent".
78
"pref" vs "pre'fe'rence" is best scored as the first "'" being dropped
79
and the trailing "e'rence" also being dropped. 10 points for the accent
80
drop plus 7 for the trailing "e'rence". Total of 17 points.
82
"pref" vs "pre'ce'dent" seems like it should be scored worse since the
83
accent is dropped and the following character is incorrect. Instead,
84
however, the algorithm assigns a lower score for the possibility that
85
the accent is replaced by the letter "f" (10 points) and then the rest
86
of the characters "ce'dent" are dropped (7 points). Also a total of 17
89
This tie would be prevented with a higher swap cost.
91
This problem could be fixed by special-casing the dropping of accents in
92
the core search algorithm and I tested this approach but it had a
93
substantial performance cost on the rather tight inner loop (nearly
94
doubling the amount of time to do a search).
96
Another possibility would be to drop all accents entirely during the
97
normalisation process. This sounds acceptable for French, but I imagine
98
other languages may find this to be less than ideal.
100
Also, consider "fl" against "file" and "fix" as another example that
101
doesn't rely on accents to make its case (although the argument is
102
somewhat weaker here). This shows that we should probably just increase
105
2012-04-02 Ryan Lortie <desrt@desrt.ca>
107
hud-performance: increase the size of the test again
109
The testcase size was reduced for easier valgrinding. That accidentally
110
got committed with a fix for a leak that was found while valgrinding.
111
Put it back to the larger value.
113
2012-04-02 Ryan Lortie <desrt@desrt.ca>
115
'make distcheck' fixes
117
- fix srcdir != builddir issue for new testcase
118
- make sure the new gsettings schemas get disted
120
2012-04-02 Ryan Lortie <desrt@desrt.ca>
122
fix small leak in testcase
124
2012-04-02 Ryan Lortie <desrt@desrt.ca>
126
really. don't delete schemas on 'make clean'.
128
2012-04-01 Ryan Lortie <desrt@desrt.ca>
130
Skip checking of search terms when it's not possible to take the term
132
There are two situations in which a term will never be taken:
134
- the max cost has already been exceeded
136
- the cost of taking the term can't possibly beat the cost of dropping it
138
In those two cases, don't do the extra work.
140
2012-04-01 Ryan Lortie <desrt@desrt.ca>
142
Add a good search performance test
144
2012-03-31 Ryan Lortie <desrt@desrt.ca>
146
test that the new distance algorithm fixes bug 947283
148
The new needle/haystack matching algorithm penalises trailing haystack
149
terms that are not matched to needles. Result: "save" now matches "File
150
> Save" better than it matches "File > Save As...".
152
2012-03-31 Ryan Lortie <desrt@desrt.ca>
154
don't blow away GSettings schemas on 'make clean'
156
2012-03-31 Ryan Lortie <desrt@desrt.ca>
160
2012-03-30 Ryan Lortie <desrt@desrt.ca>
162
tests: test-distance: write compatibility shim
164
Write a compatibility shim for the old calculate_distance() API so that
165
we can use the existing testcases to test the new code for regressions.
167
2012-03-30 Ryan Lortie <desrt@desrt.ca>
169
Don't leak the app icon.
171
The app icon was leaking once per menu item. This could get pretty substantial
174
2012-03-30 Ryan Lortie <desrt@desrt.ca>
176
bring back the results highlighting
178
2012-03-30 Ryan Lortie <desrt@desrt.ca>
180
HudStringList -> HudTokenList conversion: do it in the right order
182
The current algorithm was resulting in "File > Save" turning into the
183
token list "Save", "File", which is problematic given the
184
order-sensitive nature of the new assignment algorithm.
186
2012-03-30 Ryan Lortie <desrt@desrt.ca>
188
Remove exponential algorithm for matching haystack and needle tokens
190
Instead, use a new O(n^2) algorithm inspired by the Levenshtein distance.
192
2012-03-30 Ryan Lortie <desrt@desrt.ca>
196
Drop the no-longer used settings. Improve the documentation for the
199
Change the default end-drop penalty from 10 to 1.
201
Rename "drop-penalty-end" to "end-drop-penalty" for consistency (all
202
other penalties have the form '*-penalty') and also because the default
203
value changed and we don't want to accidentally pick up any old value
204
the user may have around.
206
2012-03-30 Ryan Lortie <desrt@desrt.ca>
208
Switch to using new intltool gschema support.
210
Drop the .gschema.xml.in dance.
212
2012-03-30 Ryan Lortie <desrt@desrt.ca>
214
Simplify the improve the performance of the distance calculation.
216
- we now normalise to casefolded normal-form ucs4 ahead of time and
217
operate on that instead of doing the normalisation at each search.
218
This removes normalisation steps and utf8 operations from the main
221
- the innermost loop has been simplified free it of special-cases: it
222
is now exactly equal to the reference implementation of Levenshtein
223
distance from Wikipedia
225
- we continue to support post-drops as a special-case, but other
226
special cases have been removed. pre-adds were probably a bad idea
227
to start with, but it's worth talking about bringing back transposes.
229
- 'bold' highlighting of matching search terms is broken again (and
232
2012-04-03 Charles Kerr <charles.kerr@canonical.com>
234
merge p:~charlesk/indicator-appmenu/lp-957417 to silence a coverity warning by removing a goto.
236
2012-04-03 Charles Kerr <charles.kerr@canonical.com>
238
remove a goto to silence a Coverity warning reported in lp bug #957417
240
2012-03-29 Charles Kerr <charles.kerr@canonical.com>
242
Merge lp:~desrt/indicator-appmenu/simplify-math to perform some explicit strength reduction to clarify an algorithm. Also, use a signed integer to correct an error due to unsigned substraction.
244
2012-03-29 Ryan Lortie <desrt@desrt.ca>
246
Perform some explicit strength reduction to clarify an algorithm
248
Also, use a signed integer to correct an error due to unsigned substraction.
250
2012-03-29 Charles Kerr <charles.kerr@canonical.com>
252
merge lp:~ken-vandine/indicator-appmenu/tests_use_builddir to use mock-json-app in the builddir instead of depending on an installed version of indicator-appmenu-tools, which caused a bootstrapping problem
254
2012-03-29 Ken VanDine <ken.vandine@canonical.com>
256
Use mock-json-app in the builddir instead of depending on an installed version of indicator-appmenu-tools, which causes a bootstrapping problem
3
258
2012-03-28 Charles Kerr <charles.kerr@canonical.com>