/
githubmirror
/
cmssw
Обзор
Документация
Войти
/
githubmirror
/
cmssw
Код
Запросы
0
Пакеты
0
Релизы
0
Аналитика
Безопасность
master
FWCore/MessageService/test/AAA_AtlasOfTests
198 строк
8 KB
Christopher Jones
Changed category FwkJob to FwkTest for all tests
20 окт 2020, 16:32
20 окт 2020, 16:32
3d26a59
Код
Авторство
О чём код?
Atlas of MessageLogger tests Tests marked $ are tested in both single thread and multithread running. Tests marked % are tested with both the default and EDM_ML_DEBUG defined. u1 $% Gives different thresholds to multiple destinations, and assigns limit=0 to annoying message categories preEventProcessing and FwkTest. Incidnetally tests noTimeStamps. Also generates a job report file (u1_job_report.mxml). ---------- UnitTestClient_A ---------- UnitTestClient_Ad u2 $ Tests handling of a message generated pre-configuration, by using the special generate_preconfiguration_message parameter in the cfg file. ---------- UnitTestClient_A u3 Generates a plain-vanilla statistics destination (not matching any output destination). ---------- UnitTestClient_A u4 Generates a pair of statistics destinations but assigns a different file name for one of them (other than the pset name) and a threshold for the other. ---------- UnitTestClient_A u5 $ Tests reset versus no reset behavior when multiple statistics summaries are triggered. (That is, if untracked bool reset = true then the next time statistics are triggered, they start from having been cleared the last time the sumary was sent out.) ---------- UnitTestClient_B u6 $ Tests hex and other manipulators ---------- UnitTestClient_C u7 Filtering all but one category (for example FwkTest). Creates a fwkJobReports u7_job_report.mxml as well as logs. ---------- UnitTestClient_C u8 $ Tests defaults and limits: overall defaults of no time stamps overall defaults for all destinations, for unnamed categories, and for specific category default limit for a destination superceding overall default default limit for a destination not superceding specific category overall default limit for specific category superceding both defaults ---------- UnitTestClient_E u9 $ Tests behavior when using the official MessageLogger.cfi. ---------- UnitTestClient_G u10 Use of extensionless filenames ---------- UnitTestClient_C u11 $ reportEvery, and some overall defaults overall defaults regarding reportEvery overall defaults for all destinations, for unnamed categories overall defaults for all destinations, specific category default reportEvery for a destination superceding overall default default reportEvery for a destination not superceding specific category overall default reportEvery for specific category superceding both defaults reportEvery and limits interacting with each other ---------- UnitTestClient_F u12 Tests the placeholder feature. ---------- UnitTestClient_A u13 Tests LogVerbatim and LogTrace ---------- UnitTestClient_H u14 $ Message suppression by severity and by module. For example, suppressInfo = { "sendSomeMessages", "source" } Does not test suppressInfo = "*" or any excludeModule or such. ---------- UnitTestClient_I u15 LogVerbatim and LogTrace; also whether isInfoEnalbled() works when it is enabled. ---------- UnitTestClient_H u16 $ Explicit specification of filenames and of extensions in destinations, statistics, and jobReports. Debug is enabled. ---------- UnitTestClient_Ad u17 Threshold levels for destinations, limit=0 for a category, enabling all (*) LogDebug, with one destination responding. Also verify that by default, the threshold for a destination is INFO, and verify name used for "severe" errors is System, not Severe, in summary statistics, and jobReports. ---------- UnitTestClient_K u18 LogSystem, LogAbsolute, LogProblem, LogPrint, and LogVerbatim. ---------- UnitTestClient_M u19 % Long header lines, to test the non-breaking behavior. ---------- UnitTestClient_N u20 $ Test default configuration file for MessageLogger service for users that include MessageLogger.cfi and nothing else. Side-tests the behavior of isMessageProcessingSetUp() when it should be returning true. ---------- UnitTestClient_G u21 By-severity default limit on each type of message, and specific-category override of that default. ---------- UnitTestClient_O u22 $ Tests effect of LogFlush by cfg-configurable choices of how many messages to use to clog the queue and whether or not FlushMessageLog is invoked. Under normal testing, it will invoke FlushMessageLog in a situation where its absence would result in different output. ---------- UnitTestClient_P u23 $ Tests effect of GroupLogStatistics(category) by having 6 differently labeled modules (of two distinct classes) all of which issue LogIssue with two "grouped" categories. ---------- UnitTestClient_Q u24 Behavior when duplicate file names are supplied. We create two destinations, with ERROR and WARNING thresholds, and force the files both to be u24.log. ---------- UnitTestClient_A u25 Imposition of limits and thresholds of specific interest to user requests: Limits per-category on ERROR and WARNING messages. ---------- UnitTestClient_R u27 $ Tests proper output of both ungrouped and grouped categories in summary placed in the FrameworkJobReport.xml. ---------- UnitTestClient_Q u28 Tests distinct threshold level for linked statistics destination: A destination with different PSet name uses output= to share the same stream as an ordinary output destination. But their thresholds are different. The statistics destination does ot report the INFO messages that the output destination did get. ---------- UnitTestClient_A u30 $ Tests LoggedErrorsSummary, using modues that have normal categories and also grouped categories. The expectation is that even grouped categories will be separated by module on the per-event summary. ---------- UnitTestClient_S u31 $ Tests LoggerErrorsSummary() and LoggerErrorsOnlySummary() ---------- UnitTestClient_T u33 $% Tests behavior when modules log outside the per-event analyze Tests that each type of activity gets right module label Checks suppression and enabling of severities and LogDebug These tests are done in two versions - t (threads on) and not t ---------- UnitTestClient_V ---------- UnitTestClient_Vd u34 Tests that when no destinations have threshold as low asINFO, LogInfo is suppressed (rather than just ignored by all destinations). Uses a special testing class LogWarningThatSuppressesLikeLogInfo. ---------- UnitTestClient_W u35 Tests that when any destinations has threshold as low asINFO, LogInfo is not suppressed. Other half of test infvolving u34. Uses a special testing class LogWarningThatSuppressesLikeLogInfo. ---------- UnitTestClient_W Non-regression-suite tests (not run via scramv1 b runtests): u0 Includes the cfi file, but nothing else. ---------- UnitTestClient_G u26 Is a near-identical clone of u17, using a different client, and is not in the Buildfile ---------- UnitTestClient_A u28_cerr Tests distinct threshold level for linked statistics destination: A destination with different PSet name uses output=cerr to share the same stream as an ordinary output destination to cerr. But their thresholds are different. ---------- UnitTestClient_A u29 Tests the hardwired defaults. Does not include MessageLogger.cfi nor explicitly mention MessageLogger or MessageService at all. Unlike u20, this is not directly suitable for a unit test because the we can't disable time stamps and still be true to the "no Message ---------- UnitTestClient_A timing_t Tests output of Timing service to JobReport (Inherently tough for regression testing as times will change and cannt be suppressed.) ** Run via cmsRun -e timing_t.cfg ** Look for and look at FrameworkJobReport.xml and timing_t.log. --------- UnitTestClient_A u1 u1 can be used to verify that suppressing LogDebug quaffs any calls made on the LogDebug line, even if they have a side effect, by switching in a different unit test that makes such a call. ---------- UnitTestClient_A_suppTest.cc o1 JobReport output of Memory service