Spell checking API tests are failing on the bots: [ RUN ] EWK2UnitTestBase.ewk_text_checker_string_spelling_check_cb_set /home/buildslave-1/webkit-buildslave/efl-linux-64-debug-wk2/build/Source/WebKit2/UIProcess/API/efl/tests/test_ewk2_text_checker.cpp:150: Failure Value of: text Actual: "aa " Expected: expectedMisspelledWord Which is: "aa" /home/buildslave-1/webkit-buildslave/efl-linux-64-debug-wk2/build/Source/WebKit2/UIProcess/API/efl/tests/test_ewk2_text_checker.cpp:468: Failure Value of: callbacksExecutionStats.spellingCheck Actual: false Expected: true [ FAILED ] EWK2UnitTestBase.ewk_text_checker_string_spelling_check_cb_set (209 ms) [ RUN ] EWK2UnitTestBase.ewk_text_checker_word_guesses_get_cb_set /home/buildslave-1/webkit-buildslave/efl-linux-64-debug-wk2/build/Source/WebKit2/UIProcess/API/efl/tests/test_ewk2_text_checker.cpp:150: Failure Value of: text Actual: "aa " Expected: expectedMisspelledWord Which is: "aa" [ FAILED ] EWK2UnitTestBase.ewk_text_checker_word_guesses_get_cb_set (211 ms) [ RUN ] EWK2UnitTestBase.ewk_text_checker_word_learn_cb_set /home/buildslave-1/webkit-buildslave/efl-linux-64-debug-wk2/build/Source/WebKit2/UIProcess/API/efl/tests/test_ewk2_text_checker.cpp:150: Failure Value of: text Actual: "aa " Expected: expectedMisspelledWord Which is: "aa" [ FAILED ] EWK2UnitTestBase.ewk_text_checker_word_learn_cb_set (210 ms) [ RUN ] EWK2UnitTestBase.ewk_text_checker_word_ignore_cb_set /home/buildslave-1/webkit-buildslave/efl-linux-64-debug-wk2/build/Source/WebKit2/UIProcess/API/efl/tests/test_ewk2_text_checker.cpp:150: Failure Value of: text Actual: "aa " Expected: expectedMisspelledWord Which is: "aa" [ FAILED ] EWK2UnitTestBase.ewk_text_checker_word_ignore_cb_set (195 ms)
Hi, As we discussed on irc, to do not break WK2-EFL's unit tests we shouldn't call callback checkSpellingOfString for the word separators explicitly, for example when WebCore requests check spelling for "zz " it means that we should call callback function with "zz " only, not for "zz " and " ". To achieve this we may skip word separators, using textBreak* methods (not possible to just skip white spaces as it doesn't solve this issue for CJK). My suggestion is to rollout bug 107682 as the unit tests have shown that unified text checker is not fully implemented. We may consider to skip spelling unit tests until bug 108299 lands. Feel free to choose the most suitable option for you.
Is it really a bug that the callback is called with "zz " and then " "? I had a patch to make the EFL API tests handle this properly.
*** This bug has been marked as a duplicate of bug 108349 ***