Mercurial > vim
view src/testdir/runtest.vim @ 12905:30b0d784c870 v8.0.1329
patch 8.0.1329: when a flaky test fails it also often fails the second time
commit https://github.com/vim/vim/commit/550586015232ecc4f68b3479fa8ba2cf81c76e7b
Author: Bram Moolenaar <Bram@vim.org>
Date: Tue Nov 21 15:14:51 2017 +0100
patch 8.0.1329: when a flaky test fails it also often fails the second time
Problem: When a flaky test fails it also often fails the second time.
Solution: Sleep a couple of seconds before the second try.
author | Christian Brabandt <cb@256bit.org> |
---|---|
date | Tue, 21 Nov 2017 15:15:05 +0100 |
parents | 90aaa974594e |
children | 6cf6138b6484 |
line wrap: on
line source
" This script is sourced while editing the .vim file with the tests. " When the script is successful the .res file will be created. " Errors are appended to the test.log file. " " To execute only specific test functions, add a second argument. It will be " matched against the names of the Test_ funtion. E.g.: " ../vim -u NONE -S runtest.vim test_channel.vim open_delay " The output can be found in the "messages" file. " " The test script may contain anything, only functions that start with " "Test_" are special. These will be invoked and should contain assert " functions. See test_assert.vim for an example. " " It is possible to source other files that contain "Test_" functions. This " can speed up testing, since Vim does not need to restart. But be careful " that the tests do not interfere with each other. " " If an error cannot be detected properly with an assert function add the " error to the v:errors list: " call add(v:errors, 'test foo failed: Cannot find xyz') " " If preparation for each Test_ function is needed, define a SetUp function. " It will be called before each Test_ function. " " If cleanup after each Test_ function is needed, define a TearDown function. " It will be called after each Test_ function. " " When debugging a test it can be useful to add messages to v:errors: " call add(v:errors, "this happened") " Without the +eval feature we can't run these tests, bail out. so small.vim " Check that the screen size is at least 24 x 80 characters. if &lines < 24 || &columns < 80 let error = 'Screen size too small! Tests require at least 24 lines with 80 characters' echoerr error split test.log $put =error w cquit endif " Common with all tests on all systems. source setup.vim " For consistency run all tests with 'nocompatible' set. " This also enables use of line continuation. set nocp viminfo+=nviminfo " Use utf-8 or latin1 by default, instead of whatever the system default " happens to be. Individual tests can overrule this at the top of the file. if has('multi_byte') set encoding=utf-8 else set encoding=latin1 endif " Avoid stopping at the "hit enter" prompt set nomore " Output all messages in English. lang mess C " Always use forward slashes. set shellslash let s:srcdir = expand('%:p:h:h') " Prepare for calling test_garbagecollect_now(). let v:testing = 1 " Support function: get the alloc ID by name. function GetAllocId(name) exe 'split ' . s:srcdir . '/alloc.h' let top = search('typedef enum') if top == 0 call add(v:errors, 'typedef not found in alloc.h') endif let lnum = search('aid_' . a:name . ',') if lnum == 0 call add(v:errors, 'Alloc ID ' . a:name . ' not defined') endif close return lnum - top - 1 endfunc func RunTheTest(test) echo 'Executing ' . a:test " Avoid stopping at the "hit enter" prompt set nomore " Avoid a three second wait when a message is about to be overwritten by the " mode message. set noshowmode " Clear any overrides. call test_override('ALL', 0) " Some tests wipe out buffers. To be consistent, always wipe out all " buffers. %bwipe! " The test may change the current directory. Save and restore the " directory after executing the test. let save_cwd = getcwd() if exists("*SetUp") try call SetUp() catch call add(v:errors, 'Caught exception in SetUp() before ' . a:test . ': ' . v:exception . ' @ ' . v:throwpoint) endtry endif call add(s:messages, 'Executing ' . a:test) let s:done += 1 if a:test =~ 'Test_nocatch_' " Function handles errors itself. This avoids skipping commands after the " error. exe 'call ' . a:test else try exe 'call ' . a:test catch /^\cskipped/ call add(s:messages, ' Skipped') call add(s:skipped, 'SKIPPED ' . a:test . ': ' . substitute(v:exception, '^\S*\s\+', '', '')) catch call add(v:errors, 'Caught exception in ' . a:test . ': ' . v:exception . ' @ ' . v:throwpoint) endtry endif if exists("*TearDown") try call TearDown() catch call add(v:errors, 'Caught exception in TearDown() after ' . a:test . ': ' . v:exception . ' @ ' . v:throwpoint) endtry endif " Clear any autocommands au! " Close any extra tab pages and windows and make the current one not modified. while tabpagenr('$') > 1 quit! endwhile while 1 let wincount = winnr('$') if wincount == 1 break endif bwipe! if wincount == winnr('$') " Did not manage to close a window. only! break endif endwhile exe 'cd ' . save_cwd endfunc func AfterTheTest() if len(v:errors) > 0 let s:fail += 1 call add(s:errors, 'Found errors in ' . s:test . ':') call extend(s:errors, v:errors) let v:errors = [] endif endfunc " This function can be called by a test if it wants to abort testing. func FinishTesting() call AfterTheTest() " Don't write viminfo on exit. set viminfo= " Clean up files created by setup.vim call delete('XfakeHOME', 'rf') if s:fail == 0 " Success, create the .res file so that make knows it's done. exe 'split ' . fnamemodify(g:testname, ':r') . '.res' write endif if len(s:errors) > 0 " Append errors to test.log split test.log call append(line('$'), '') call append(line('$'), 'From ' . g:testname . ':') call append(line('$'), s:errors) write endif let message = 'Executed ' . s:done . (s:done > 1 ? ' tests' : ' test') echo message call add(s:messages, message) if s:fail > 0 let message = s:fail . ' FAILED:' echo message call add(s:messages, message) call extend(s:messages, s:errors) endif " Add SKIPPED messages call extend(s:messages, s:skipped) " Append messages to the file "messages" split messages call append(line('$'), '') call append(line('$'), 'From ' . g:testname . ':') call append(line('$'), s:messages) write qall! endfunc " Source the test script. First grab the file name, in case the script " navigates away. g:testname can be used by the tests. let g:testname = expand('%') let s:done = 0 let s:fail = 0 let s:errors = [] let s:messages = [] let s:skipped = [] if expand('%') =~ 'test_vimscript.vim' " this test has intentional s:errors, don't use try/catch. source % else try source % catch let s:fail += 1 call add(s:errors, 'Caught exception: ' . v:exception . ' @ ' . v:throwpoint) endtry endif " Names of flaky tests. let s:flaky = [ \ 'Test_client_server()', \ 'Test_close_and_exit_cb()', \ 'Test_collapse_buffers()', \ 'Test_communicate()', \ 'Test_exit_callback_interval()', \ 'Test_nb_basic()', \ 'Test_oneshot()', \ 'Test_pipe_through_sort_all()', \ 'Test_pipe_through_sort_some()', \ 'Test_quoteplus()', \ 'Test_quotestar()', \ 'Test_reltime()', \ 'Test_terminal_composing_unicode()', \ 'Test_terminal_noblock()', \ 'Test_terminal_redir_file()', \ 'Test_with_partial_callback()', \ ] " Locate Test_ functions and execute them. redir @q silent function /^Test_ redir END let s:tests = split(substitute(@q, 'function \(\k*()\)', '\1', 'g')) " If there is an extra argument filter the function names against it. if argc() > 1 let s:tests = filter(s:tests, 'v:val =~ argv(1)') endif " Execute the tests in alphabetical order. for s:test in sort(s:tests) " Silence, please! set belloff=all call RunTheTest(s:test) if len(v:errors) > 0 && index(s:flaky, s:test) >= 0 call add(s:messages, 'Found errors in ' . s:test . ':') call extend(s:messages, v:errors) call add(s:messages, 'Flaky test failed, running it again') let first_run = v:errors " Flakiness is often caused by the system being very busy. Sleep a couple " of seconds to have a higher chance of succeeding the second time. sleep 2 let v:errors = [] call RunTheTest(s:test) if len(v:errors) > 0 let second_run = v:errors let v:errors = ['First run:'] call extend(v:errors, first_run) call add(v:errors, 'Second run:') call extend(v:errors, second_run) endif endif call AfterTheTest() endfor call FinishTesting() " vim: shiftwidth=2 sts=2 expandtab