qa:runtests:documentation

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
qa:runtests:documentation [2009/08/21 15:36]
g2 added parallel execution doc
qa:runtests:documentation [2017/09/22 13:28] (current)
Line 3: Line 3:
  
   *[[http://wiki.php.net/qa/runtests/documentation/commandline|Command line options]]   *[[http://wiki.php.net/qa/runtests/documentation/commandline|Command line options]]
-A list of the options available in the hold version of run-tests and information about whether that are (or will be) re-implemented.+A list of the options available in the old version of run-tests and information about whether that are (or will be) re-implemented.
  
-===== Parallel Execution =====+====== Introduction ====== 
 +runtests is a test environment written in PHP. It is used to test PHP itself. 
 +It is not a replacement for unit test frameworks like PHPUnit. 
 + 
 +The basic principle of automated tests is very simple. Compare a computed 
 +result with some pre-calculated "known good", expected value. 
 +If both match, the test has passed. If they do not match, the tests fails, and 
 +there is probably a bug in PHP, or some precondition for the test was not met. 
 + 
 +Testing PHP is a far more complex issue than it may seem at first glance, 
 +because each test may require an individual PHP configuration, certain settings 
 +at operating system level (for example environment variables), or even external 
 +services (like databases, LDAP servers or IMAP servers). 
 + 
 +A test that requires external services can be complex to set up, but there is  
 +no way around that. How could you make sure that running a SQL statement against 
 +a database works without actually running it against a database? 
 + 
 +Since a failing test will probably leave its environment in an unpredictable 
 +state, a high level of test isolation is required. In other words, this means 
 +that we need to set up a new preconfigured PHP process for each test. This is  
 +the only way to ensure that every test runs in a clean environment. 
 + 
 +So, to execute each test, runtests must set up a PHP process, have it run the 
 +test, collect the output, and compare it to the pre-calculated "known good" 
 +expected output. That spawned PHP process (hopefully) terminates, so at the 
 +process level, we do not have to worry about the potential mess that a test has 
 +left behind. Still, if the test has created files, databases, or modified the 
 +global system environment, additional work may be required to clean up after 
 +the test. 
 + 
 +Since PHP runs on most of the available platforms, runtests must also run on  
 +all these platforms. From an implementation point of view, this means that 
 +runtests must work on a minimal, stock PHP installation, and should make as  
 +few as possible assumptions on the system environment. All the more or less 
 +known cross-platform issues like different line endings, directory separators,  
 +case handling, as well as limits of different operating system families 
 +(file name path length, include path length) have to be taken into account and  
 +being dealt with in runtests. 
 + 
 +Since runtests spawns off a separate PHP process to run a test, it is not 
 +necessary to run each test in the same PHP version that runtests runs on. 
 +In other words, that means that you can test a different PHP version than 
 +you are actually running - and in fact, it may be a good idea to use a 
 +"known good" PHP version to test a new, less tested PHP version. After all, 
 +bugs in the PHP version runtests itself is running on might affect runtests 
 +itself, and thus make the test results less reliable. 
 + 
 +We have already mentioned that tests need an individual environment. Not all  
 +tests can be run at the command line, for example. If a test needs to make 
 +sure that PHP returns correct HTTP header, or processes GET or POST input, 
 +the test probably requires the CGI SAPI to run. 
 + 
 +The implementation of runtests that is described in these pages is specific to PHP5.3 and beyond. Many of the implementation details are similar to the previous version of runtests but the overall code structure is very different.  
 + 
 +====== Overall code structure ====== 
 +The most important classes in classes in the code are shown in the figure below and described in the following paragraphs. 
 + 
 +{{:qa:runtests:mainclasses.png|}} 
 + 
 +===== Test Run ===== 
 +The main class is rtPhpTestRun. This class is responsible for the overall running of one or more test cases. 
 + 
 +rtPhpTestRun instantiates the run configuration class (rtRuntestsConfiguration, see Configuration) and then executes a single test or a group of tests. 
 +===== Test Group ===== 
 +A test group (rtPhpTestGroup) is currently defined as all of the tests within a single directory.  It seems likely that groups of tests can be run in parallel, teh prototype code is designed to enable that. 
 + 
 +There is currently no group configuration class, however, it is possible that one may be required. For example there might be groups of tests that cannot be run at the same time as other groups (tests with REDIRECT?).  
 +===== Test Case ===== 
 +Each test case is executed in it's own process. Many classes are associated with the execution of a single test case. The following subsections give a brief description of the responsibilities of each class. 
 + 
 +===== Run configuration ===== 
 +This group of classes is responsible for setting the configuration for the whole test run. So, for example the name of the name of the PHP executable under test is set in this group as is the name of the PHP CGI exectuable. 
 + 
 +===== Test case configuration ===== 
 +Some settings are specific to each test case and need to be set for a single test. These classes are responsible for settings at the level of individual tests. 
 + 
 +====== Run configuration classes ====== 
 + 
 +This sections contains a brief description of each class involved in setting the configuration for the whole test run. 
 + 
 +==== rtRunTestsConfiguration ==== 
 + 
 +The main class in this section - all of the other classes are related to setting options in the run configuration object. There should be OS dependent instances for unix, windows etc.  
 + 
 +==== rtCommandLineOptions ==== 
 +Instantiated by rtRuntestsConfiguration. Parses the command line options (argv[]) for runt-tests.php 
 + 
 +==== rtEnvironmentVariables ==== 
 +Get a list of all of the Environment variable and maintains as an array. This is required both establishing configuration settings and to pass to to proc_open() when PHP code is run as part of a test (see rtPhpRunner and  testcase/sections/executablesections). 
 + 
 +==== rtAddToCommandLine ==== 
 +Somewhat bizarrely (to my mind) argv[] for run-tests can be added to using an Environment variable (TEST_PHP_ARGS). This class is instantiated after rtEnviromentVariables and may (if TEST_PHP_ARGS is used) add other options to an instance of rtCommandLineOptions. This is here for compatibility with the old version of run-tests, if no good reason for it comes to light it should go. 
 + 
 +==== rtIniAsCommandLineArgs ==== 
 +The run-tests code pre-sets a number of  ini settings automatically, for example  error_reporting. This helps prevent users accidentally picking up different php.ini files which would cause tests to fail. This class maintains a list of ini settings and converts them to -d flags used in teh PHP command line when a test case is run, eg 
 + 
 +php_exectuable   -d flag1 -d flag2...   test.php   test_arguments 
 + 
 +Note that these can be added to using the test case section INI. This class is instantiated from the setting class that builds the php command line, it is also instantiated from the rtTestConfiguration class when a test case contains an INI section. 
 +rtPreCondition list 
 +Maintains a list of all of the pre-conditions that must be satisfied before a test run is attempted. For example - check that a PHP executable has been specified. 
 + 
 +==== rtPrecondition ==== 
 +Parent class for the classes that contain code to check for each pre-condition. See subdirectory 'precoditions' for the sub-classes that perform that checks. 
 + 
 +==== rtSetting ==== 
 +Parent class for classes that set run configuration settings. See 'setting' directory for sub-classes. Note that run settings can be derived from eith command line options or Environment variables, the functions of teh settings classes is mainly to determine which of these has been used and to set the run configuration accordingly. 
 + 
 +===== Test case classes ===== 
 +====rtPhpTestFile==== 
 +Reads the contents of a test file, checks that it's a valid test (see precondtions) and removes windows-style line endings. 
 +rtPhpTest 
 +The main test case class. Takes the contents of the file read by rtPhpTestFile and creates section objects for each section (see Test Case Sections). The run() method runs any executable sections and then compares teh test out put with the expected output.  
 + 
 +====rtTestPreCondtion==== 
 +There are a number of preconditions that must be met before a test can be run. This is the parent class for preconditions. Preconditions are checked as the test file is read (see section on test precondtions) 
 +====rtTestConfiguration ==== 
 +The run configuration can be modified by test case sections. The test configuration class implements the modifications and is used by rtFileSection when running the test. 
 +====rtPhpRunner ==== 
 +As the name suggests, this runs PHP code. It's an almost exact copy of the original run-tests code (system_with_timeout()). No reason to change something that works fine as it is. 
 +==== rtPhpTestResults ==== 
 +Collates the results of running (or attempting to run) a test. If the test has failed ensures that the difference between the expected and actual output is calculated and saved. Decides which files to keep and which to delete. Keeps a record of any saved files and of teh overall status (pass, fail, skip etc) of the test. 
 + 
 +==== rtTestDifference ==== 
 +Exact copy of the differencing code in the original run-tests. This code is complex, it might be possible to do better but for now it seems best not to try and modify it. 
 + 
 +==== rtOutputWriter ==== 
 +Takes the list of test results and writes in the desired format. Is a parent class for various different formats out output class. Currently only the 'list' format is implemented. 
 + 
 + 
 + 
 +====Test Section Classes ==== 
 + 
 +Each test is made up from a number of sections, for example,  
 +--TEST-- 
 +a title 
 +--FILE-- 
 +<?php 
 + some php code 
 +?> 
 +--EXPECT-- 
 +the expected result of running the PHP code.  
 + 
 +All sections have a name and some contents. Sometimes the contents just have information - for example the test title, sometimes the contents are executable, or used to compare against the output. 
 + 
 +The base class for all sections is rtSection. This class has a method which initialises the contents to an array of strings. 
 + 
 +There are four subclasses which extend rtSection, these are rtInformationSection, rtExecutableSection, rtoutputSection and rtConfigurationSection. Each of these are extended again by classes which represent specific sections. The class hierarchy is shown in the attached ODP charts.  
 + 
 +A brief description of the responsibilities of each class is given below. 
 + 
 +====Information Sections==== 
 +===rtTestSection === 
 +This class just contains the test case title. 
 + 
 +===rtCreditsSection === 
 +This class contains information about who wrote the test 
 + 
 +===rtXfailSection=== 
 +The presence of this section in the test indicates that the test is expected to fail. The contents contain information about the reason for an expected failure. 
 + 
 +====Executable Sections==== 
 + 
 +===rtFileSection=== 
 +This is the most complex of the sections. The contents represent the PHP code that needs to be run. The section is responsible for assembling the command line that will run the code. The format of the command line is: 
 + 
 +php_executable    php_arguments    test.php   test_arguments 
 + 
 +Note that both php_arguments aand test_arguments can be modified by other test sections and are therefore supplied from a TestConfiguration object and not from the RunConfiguration. 
 +The executable code is supplied in a file called test.php, where 'test' is the name of the testcase. The file is constructed fron teh test case contents and written by this section. 
 +===rtSkipIfSection=== 
 +Executes code to determine if the test should be run or not. For example: 
 +<?php  
 +if( substr(PHP_OS, 0, 3) == 'WIN') {  
 +  die('skip Not for Windows');  
 +Note that the first word in the comment is always 'skip' or 'warn'. For example in ext/standard/tests/time/001.phpt: 
 + 
 +<?php  
 +        if (!function_exists('microtime'))   
 + die('skip microtime() not available');  
 +        die('warn system dependent');  
 +?>  
 +Note: the warning is a pretty strange thing to do as part of a section called SKIPIF. I've implemented it here for compatibility but it would be better to have a separate WARN section 
 + 
 + 
 + 
 +===rtCleanSection=== 
 +This section runs code that should clean up after a test. Many people forget that this is a separate piece of executable code - therefore you can't define $myFile in the body of the test and expect to be able to unling $myFile in the CLEAN section. There are currently 71 tests that need to be fixed because the CLEAN sections are a mess. 
 + 
 +The old run-tests code just chucks away any results from running a CLAEN section. In this code if the results of running the clean section are not an empty string then the results are printed out with a WARN flag. 
 +====OutputSections==== 
 +These sections deal with the expected output of the test and compare it with the actual output. 
 +rtExpectSection 
 + 
 +rtExpectFSection 
 + 
 +rtExectRegexSection 
 + 
 +====Configuration sections==== 
 +These sections have additional configuration information that changes the way that a  single test is run 
 +===rtArgsSection=== 
 +This section contains command line arguments for the test case.  For an example of the usage see 
 +ext/standard/tests/general_functions/getopt.phpt 
 + 
 +Test test_arguments are appended after the testcase as a string preceeded by '--' 
 + 
 +php_executable    php_arguments    test.php    test_arguments 
 +===rtEnvSection=== 
 +This section adds any environmental variables that are required by a single testcase. For an example of usage see: 
 +ext/standard/tests/general_functions/parse_ini_basic.phpt  
 +The environmental variables specified in the test case are appended to the array  of environment variables which is constructed in the run configuration and set in teh testsConfiguration 
 +===rtIniSection=== 
 +This section adds additional commanline arguments to the PHP command line: 
 +php_executable    php_arguments    test.php    test_arguments 
 + 
 +These are added in the form of -d flags and are appended to those already set in the runConfiguration 
 + 
 +====== Parallel Execution ======
  
 ==== Introduction ==== ==== Introduction ====
  
-The parallel execution of the test cases is one of the main features of the new phpruntests. Compared to the previous run-tests script it allows to run the tests cases in about half the time. Before digging into the details there are a few issues to know: +The parallel execution of the test cases is one of the main features of the new phpruntests. Compared to the previous run-tests script it allows to run the tests cases in about half the time. Here are a few data about the performance: 
 + 
 +  *[[http://wiki.php.net/qa/runtests/documentation/performance|phpruntests performance]] 
 + 
 +Before digging into the details there are a few issues to know: 
  
 === Preconditions === === Preconditions ===
Line 64: Line 286:
   * resultList - array - the result list in the same order as the taskList   * resultList - array - the result list in the same order as the taskList
   * processCount - int - the number of processes (default=0). If the value is 0, the tasks are automatically executed sequentially.    * processCount - int - the number of processes (default=0). If the value is 0, the tasks are automatically executed sequentially. 
-  * reportStatus - int - defines the detail-level of the output during the test execution: +  * reportStatus - int - defines the detail-level of the output during the test execution (see {{http://wiki.php.net/qa/runtests/documentation#verbose-mode_output|verbose-mode}})
- (-1): no output +
- 0: a dot per test case +
- 1: basic information (test cases and status) +
- 2: basic information plus additional information about not passed tests +
- 3: every available information about the test case is shown +
  
 {{:qa:runtests:rttaskscheduler.png|}} {{:qa:runtests:rttaskscheduler.png|}}
qa/runtests/documentation.1250868976.txt.gz · Last modified: 2017/09/22 13:28 (external edit)