- I wrapped this in a twisty because it is not particularly relevant for the work that remains -- MichaelTempest - 03 Jul 2010
There is something bizarre going on with perl's Encode module. The SeleniumConfigTests run fine on their own and also in conjunction with some other tests. But (for example) running the AccessControlTests and SeleniumConfigTests together makes Encode produce an error (unknown encoding 'utf8') during cleanup, when WWW::Selenium::DESTROY calls Encode::decode_utf8.
The SeleniumConfigTests run fine when executed in the same test run after each of these:
AttrsTests, AutoAttachTests, ClientTests, ConfigureTests, ExampleTests, ExceptionTests,
ExpandMacrosTests, Fn_FOREACH, Fn_FORMFIELD, Fn_ICON, Fn_QUERYPARAMS, Fn_WEBLIST, FormattingTests,
FoswikiPmFunctionsTests, HierarchicalWebsTests, InitFormTests, ManageDotPmTests, MetaTests, NetTests,
PasswordTests, PluginHandlerTests, PrefsTests, RenderFormTests, RequestCacheTests, ResponseTests,
SaveScriptTests
The SeleniumConfigTests run fine when executed in the same test run before each of these:
StoreSmokeTests, StoreTests, TOCTests, TemplatesTests, TimeTests, UIFnCompileTests, UTF8Tests,
UploadScriptTests, VariableTests, ViewFileScriptTests, ViewParamSectionTests, ViewScriptTests
Anomalies:
- unknown encoding 'utf8' error - AccessControlTests, Fn_ENCODE, Fn_GROUPINFO, Fn_IF, Fn_INCLUDE, Fn_QUERY, Fn_REVINFO, Fn_SEARCH, Fn_SECTION, Fn_URLPARAM, Fn_USERINFO, FormDefTests, FuncTests, FuncUsersTests, MergeTests, RcsTests, RegisterTests, RenameTests, RobustnessTests
- Fail when run on their own - ExtenderTests
- Fail - SemiAutomaticTestsCaseTests
- Compile error - QueryTests
- printed out additional OopsException messages when run on their own (but the tests pass) - ManageDotPmTests
- browser window did not close after tests completed - Fn_GROUPS, Fn_NOP, Fn_SCRIPTURL, Fn_SEP, Fn_TOPICLIST, RESTTests (most are very short tests)
- Corrupts
$Foswiki::cfg{DefaultUrlHost}
- RequestTests
--
MichaelTempest - 07 Mar 2010
Progress! The unknown encoding 'utf8' error is now fixed and the browser windows close reasonably reliably. I am now starting on real browser-in-the-loop tests.
--
MichaelTempest - 19 Apr 2010