0
0
mirror of https://github.com/vim/vim.git synced 2025-07-04 23:07:33 -04:00
vim/runtime/syntax/testdir/runtest.vim

613 lines
18 KiB
VimL
Raw Normal View History

" Runs all the syntax tests for which there is no "done/name" file.
"
" Current directory must be runtime/syntax.
" needed because of line-continuation lines
set cpo&vim
" Only do this with the +eval feature
if 1
" Remember the directory where we started. Will change to "testdir" below.
let syntaxDir = getcwd()
let s:messagesFname = fnameescape(syntaxDir .. '/testdir/messages')
let s:messages = []
" Erase the cursor line and do not advance the cursor. (Call the function
" after each passing test report.)
def EraseLineAndReturnCarriage(line: string)
const full_width: number = winwidth(0)
const half_width: number = full_width - (full_width + 1) / 2
if strlen(line) > half_width
echon "\r" .. repeat("\x20", full_width) .. "\r"
else
echon repeat("\x20", half_width) .. "\r"
endif
enddef
" Add one message to the list of messages
func Message(msg)
echomsg a:msg
call add(s:messages, a:msg)
endfunc
" Report a fatal message and exit
func Fatal(msg)
echoerr a:msg
call AppendMessages(a:msg)
qall!
endfunc
" Append s:messages to the messages file and make it empty.
func AppendMessages(header)
silent exe 'split ' .. s:messagesFname
call append(line('$'), '')
call append(line('$'), a:header)
call append(line('$'), s:messages)
let s:messages = []
silent wq
endfunc
" Relevant messages are written to the "messages" file.
" If the file already exists it is appended to.
silent exe 'split ' .. s:messagesFname
call append(line('$'), repeat('=-', 70))
call append(line('$'), '')
let s:test_run_message = 'Test run on ' .. strftime("%Y %b %d %H:%M:%S")
call append(line('$'), s:test_run_message)
silent wq
if syntaxDir !~ '[/\\]runtime[/\\]syntax\>'
call Fatal('Current directory must be "runtime/syntax"')
endif
if !isdirectory('testdir')
call Fatal('"testdir" directory not found')
endif
" Use the script for source code screendump testing. It sources other scripts,
" therefore we must "cd" there.
cd ../../src/testdir
source screendump.vim
exe 'cd ' .. fnameescape(syntaxDir)
" For these tests we need to be able to run terminal Vim with 256 colors. On
" MS-Windows the console only has 16 colors and the GUI can't run in a
" terminal.
if !CanRunVimInTerminal()
call Fatal('Cannot make screendumps, aborting')
endif
cd testdir
if !isdirectory('done')
call mkdir('done')
endif
if !isdirectory('failed')
call mkdir('failed')
endif
set nocp
set nowrapscan
set report=9999
set modeline
set debug=throw
set nomore
au! SwapExists * call HandleSwapExists()
func HandleSwapExists()
" Ignore finding a swap file for the test input, the user might be editing
" it and that's OK.
if expand('<afile>') =~ 'input[/\\].*\..*'
let v:swapchoice = 'e'
endif
endfunc
" Trace ruler liveness on demand.
if !empty($VIM_SYNTAX_TEST_LOG) && filewritable($VIM_SYNTAX_TEST_LOG)
def s:TraceRulerLiveness(context: string, times: number, tail: string)
writefile([printf('%s: %4d: %s', context, times, tail)],
$VIM_SYNTAX_TEST_LOG,
'a')
enddef
else
def s:TraceRulerLiveness(_: string, _: number, _: string)
enddef
endif
" See ":help 'ruler'".
def s:CannotSeeLastLine(ruler: list<string>): bool
return !(get(ruler, -1, '') ==# 'All' || get(ruler, -1, '') ==# 'Bot')
enddef
def s:CannotDumpNextPage(buf: number, prev_ruler: list<string>, ruler: list<string>): bool
return !(ruler !=# prev_ruler &&
len(ruler) == 2 &&
ruler[1] =~# '\%(\d%\|\<Bot\)$' &&
get(term_getcursor(buf), 0) != 20)
enddef
def s:CannotDumpFirstPage(buf: number, _: list<string>, ruler: list<string>): bool
return !(len(ruler) == 2 &&
ruler[1] =~# '\%(\<All\|\<Top\)$' &&
get(term_getcursor(buf), 0) != 20)
enddef
def s:CannotDumpShellFirstPage(buf: number, _: list<string>, ruler: list<string>): bool
return !(len(ruler) > 3 &&
get(ruler, -1, '') =~# '\%(\<All\|\<Top\)$' &&
get(term_getcursor(buf), 0) != 20)
enddef
" Poll for updates of the cursor position in the terminal buffer occupying the
" first window. (ALWAYS call the function or its equivalent before calling
" "VerifyScreenDump()" *and* after calling any number of "term_sendkeys()".)
def s:TermPollRuler(
CannotDumpPage: func, # (TYPE FOR LEGACY CONTEXT CALL SITES.)
buf: number,
in_name_and_out_name: string): list<string>
# Expect defaults from "term_util#RunVimInTerminal()".
runtime(syntax-tests): Allow for folded and wrapped lines in syntax test files The current implementation falls short for syntax test files on two accounts: 1. With folded lines -- some lines before folded lines are unnecessarily repeated in generated dump files because closed folded lines are always treated as opened for the cursor to move _in_ instead of to move _over_ them. 2. With wrapped lines (longer than 75 columns) -- some lines are omitted in generated dump files because calculations for the cursor progress and its movement commands only refer to file lines and not their layout within a 20x75 buffer (less &cmdheight). As an alternative, we abandon deterministic (and inaccurate at times) calculations for the cursor progress and, instead, advance the cursor by as much as before for a single dump file, but now rely on marking the last visible line and additional movement to position lines at desired offsets, carefully preserving compatibility for the &scrolloff and &ruler values inherited from defaults.vim. The parent Vim process will keep track of progress through a syntax test file made by its child process ("terminal") by reading the rightmost end of the ruler line from the terminal buffer, looking for " All " or " Bot " for its cue to finish dump file generation. With these changes applied, the lossless line length limit will be raised from 75 to 1425 (for a 19x75 view) columns. Also, prefer "lastline" to "truncate" for &display; hiding the content of any last _long_ line in a view goes against the purpose of syntax file testing -- all lines should be recorded. related: #15150 fixes: #14245 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2024-05-21 01:10:26 +03:00
if winwidth(1) != 75 || winheight(1) != 20
ch_log(printf('Aborting for %s: (75 x 20) != (%d x %d)',
in_name_and_out_name,
winwidth(1),
winheight(1)))
return ['0,0-1', 'All']
runtime(syntax-tests): Allow for folded and wrapped lines in syntax test files The current implementation falls short for syntax test files on two accounts: 1. With folded lines -- some lines before folded lines are unnecessarily repeated in generated dump files because closed folded lines are always treated as opened for the cursor to move _in_ instead of to move _over_ them. 2. With wrapped lines (longer than 75 columns) -- some lines are omitted in generated dump files because calculations for the cursor progress and its movement commands only refer to file lines and not their layout within a 20x75 buffer (less &cmdheight). As an alternative, we abandon deterministic (and inaccurate at times) calculations for the cursor progress and, instead, advance the cursor by as much as before for a single dump file, but now rely on marking the last visible line and additional movement to position lines at desired offsets, carefully preserving compatibility for the &scrolloff and &ruler values inherited from defaults.vim. The parent Vim process will keep track of progress through a syntax test file made by its child process ("terminal") by reading the rightmost end of the ruler line from the terminal buffer, looking for " All " or " Bot " for its cue to finish dump file generation. With these changes applied, the lossless line length limit will be raised from 75 to 1425 (for a 19x75 view) columns. Also, prefer "lastline" to "truncate" for &display; hiding the content of any last _long_ line in a view goes against the purpose of syntax file testing -- all lines should be recorded. related: #15150 fixes: #14245 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2024-05-21 01:10:26 +03:00
endif
# A two-fold role for redrawing:
# (*) in case the terminal buffer cannot redraw itself just yet;
# (*) to avoid extra "real estate" checks.
runtime(syntax-tests): Allow for folded and wrapped lines in syntax test files The current implementation falls short for syntax test files on two accounts: 1. With folded lines -- some lines before folded lines are unnecessarily repeated in generated dump files because closed folded lines are always treated as opened for the cursor to move _in_ instead of to move _over_ them. 2. With wrapped lines (longer than 75 columns) -- some lines are omitted in generated dump files because calculations for the cursor progress and its movement commands only refer to file lines and not their layout within a 20x75 buffer (less &cmdheight). As an alternative, we abandon deterministic (and inaccurate at times) calculations for the cursor progress and, instead, advance the cursor by as much as before for a single dump file, but now rely on marking the last visible line and additional movement to position lines at desired offsets, carefully preserving compatibility for the &scrolloff and &ruler values inherited from defaults.vim. The parent Vim process will keep track of progress through a syntax test file made by its child process ("terminal") by reading the rightmost end of the ruler line from the terminal buffer, looking for " All " or " Bot " for its cue to finish dump file generation. With these changes applied, the lossless line length limit will be raised from 75 to 1425 (for a 19x75 view) columns. Also, prefer "lastline" to "truncate" for &display; hiding the content of any last _long_ line in a view goes against the purpose of syntax file testing -- all lines should be recorded. related: #15150 fixes: #14245 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2024-05-21 01:10:26 +03:00
redraw
# The contents of "ruler".
var ruler: list<string> = []
# Attempts at most, targeting ASan-instrumented Vim builds.
var times: number = 2048
# Check "real estate" of the terminal buffer. Read and compare its ruler
# line and let "Xtestscript#s:AssertCursorForwardProgress()" do the rest.
# Note that the cursor ought to be advanced after each successive call of
# this function yet its relative position need not be changed (e.g. "0%").
while CannotDumpPage(ruler) && times > 0
ruler = split(term_getline(buf, 20))
sleep 1m
times -= 1
if times % 8 == 0
redraw
endif
endwhile
TraceRulerLiveness('P', (2048 - times), in_name_and_out_name)
return ruler
enddef
" Prevent "s:TermPollRuler()" from prematurely reading the cursor position,
" which is available at ":edit", after outracing the loading of syntax etc. in
" the terminal buffer. (Call the function before calling "VerifyScreenDump()"
" for the first time.)
def s:TermWaitAndPollRuler(buf: number, in_name_and_out_name: string): list<string>
# Expect defaults from "term_util#RunVimInTerminal()".
if winwidth(1) != 75 || winheight(1) != 20
ch_log(printf('Aborting for %s: (75 x 20) != (%d x %d)',
in_name_and_out_name,
winwidth(1),
winheight(1)))
return ['0,0-1', 'All']
endif
# The contents of "ruler".
var ruler: string = ''
# Attempts at most, targeting ASan-instrumented Vim builds.
var times: number = 32768
# Check "real estate" of the terminal buffer. Expect a known token to be
# rendered in the terminal buffer; its prefix must be "is_" so that buffer
# variables from "sh.vim" can be matched (see "Xtestscript#ShellInfo()").
# Verify that the whole line is available!
while ruler !~# '^is_.\+\s\%(All\|Top\)$' && times > 0
ruler = term_getline(buf, 20)
sleep 1m
times -= 1
if times % 16 == 0
redraw
endif
endwhile
TraceRulerLiveness('W', (32768 - times), in_name_and_out_name)
if strpart(ruler, 0, 8) !=# 'is_nonce'
# Retain any of "b:is_(bash|dash|kornshell|posix|sh)" entries and let
# "CannotDumpShellFirstPage()" win the cursor race.
return TermPollRuler(
function(CannotDumpShellFirstPage, [buf, []]),
buf,
in_name_and_out_name)
else
# Clear the "is_nonce" token and let "CannotDumpFirstPage()" win any
# race.
term_sendkeys(buf, ":redraw!\<CR>")
endif
return TermPollRuler(
function(CannotDumpFirstPage, [buf, []]),
buf,
in_name_and_out_name)
runtime(syntax-tests): Allow for folded and wrapped lines in syntax test files The current implementation falls short for syntax test files on two accounts: 1. With folded lines -- some lines before folded lines are unnecessarily repeated in generated dump files because closed folded lines are always treated as opened for the cursor to move _in_ instead of to move _over_ them. 2. With wrapped lines (longer than 75 columns) -- some lines are omitted in generated dump files because calculations for the cursor progress and its movement commands only refer to file lines and not their layout within a 20x75 buffer (less &cmdheight). As an alternative, we abandon deterministic (and inaccurate at times) calculations for the cursor progress and, instead, advance the cursor by as much as before for a single dump file, but now rely on marking the last visible line and additional movement to position lines at desired offsets, carefully preserving compatibility for the &scrolloff and &ruler values inherited from defaults.vim. The parent Vim process will keep track of progress through a syntax test file made by its child process ("terminal") by reading the rightmost end of the ruler line from the terminal buffer, looking for " All " or " Bot " for its cue to finish dump file generation. With these changes applied, the lossless line length limit will be raised from 75 to 1425 (for a 19x75 view) columns. Also, prefer "lastline" to "truncate" for &display; hiding the content of any last _long_ line in a view goes against the purpose of syntax file testing -- all lines should be recorded. related: #15150 fixes: #14245 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2024-05-21 01:10:26 +03:00
enddef
func RunTest()
let XTESTSCRIPT =<< trim END
" Track the cursor progress through a syntax test file so that any
" degenerate input can be reported. Each file will have its own cursor.
let s:cursor = 1
" extra info for shell variables
func ShellInfo()
let msg = ''
for [key, val] in items(b:)
if key =~ '^is_'
let msg ..= key .. ': ' .. val .. ', '
endif
endfor
if msg != ''
echomsg msg
endif
endfunc
au! SwapExists * call HandleSwapExists()
func HandleSwapExists()
" Ignore finding a swap file for the test input, the user might be
" editing it and that's OK.
if expand('<afile>') =~ 'input[/\\].*\..*'
let v:swapchoice = 'e'
endif
endfunc
func LoadFiletype(type)
for file in glob("ftplugin/" .. a:type .. "*.vim", 1, 1)
exe "source " .. file
endfor
redraw!
endfunc
func SetUpVim()
call cursor(1, 1)
" Defend against rogue VIM_TEST_SETUP commands.
for _ in range(20)
let lnum = search('\C\<VIM_TEST_SETUP\>', 'eW', 20)
if lnum < 1
break
endif
exe substitute(getline(lnum), '\C.*\<VIM_TEST_SETUP\>', '', '')
endfor
call cursor(1, 1)
" BEGIN [runtime/defaults.vim]
" Also, disable italic highlighting to avoid issues on some terminals.
set display=lastline ruler scrolloff=5 t_ZH= t_ZR=
syntax on
" END [runtime/defaults.vim]
redraw!
endfunc
def s:AssertCursorForwardProgress(): bool
const curnum: number = line('.')
if curnum <= cursor
# Use "actions/upload-artifact@v4" of ci.yml for delivery.
writefile([printf('No cursor progress: %d <= %d (%s). Please file an issue.',
curnum,
cursor,
bufname('%'))],
'failed/00-FIXME',
'a')
bwipeout!
endif
cursor = curnum
return true
enddef
def ScrollToSecondPage(estate: number, op_wh: number, op_so: number): bool
if line('.') != 1 || line('w$') >= line('$')
return AssertCursorForwardProgress()
endif
try
set scrolloff=0
# Advance mark "c"[ursor] along with the cursor.
norm! Lmc
if foldclosed('.') < 0 &&
(strdisplaywidth(getline('.')) + &l:fdc * winheight(1)) >= estate
# Make for an exit for a screenful long line.
norm! j^
return AssertCursorForwardProgress()
else
# Place the cursor on the actually last visible line.
while winline() < op_wh
const lastnum: number = winline()
norm! gjmc
if lastnum > winline()
break
endif
endwhile
norm! zt
endif
finally
# COMPATIBILITY: Scroll up around "scrolloff" lines.
&scrolloff = max([1, op_so])
endtry
norm! ^
return AssertCursorForwardProgress()
enddef
def ScrollToNextPage(estate: number, op_wh: number, op_so: number): bool
if line('.') == 1 || line('w$') >= line('$')
return AssertCursorForwardProgress()
endif
try
set scrolloff=0
# Advance mark "c"[ursor] along with the cursor.
norm! Lmc
if foldclosed('.') < 0 &&
(strdisplaywidth(getline('.')) + &l:fdc * winheight(1)) >= estate
# Make for an exit for a screenful long line.
norm! j^
return AssertCursorForwardProgress()
else
# Place the cursor on the actually last visible line.
while winline() < op_wh
const lastnum: number = winline()
norm! gjmc
if lastnum > winline()
break
endif
endwhile
endif
finally
# COMPATIBILITY: Scroll up/down around "scrolloff" lines.
&scrolloff = max([1, op_so])
endtry
norm! zt
const marknum: number = line("'c")
# Eschew &smoothscroll since line("`c") is not supported.
# Remember that "w0" can point to the first line of a _closed_ fold
# whereas the last line of a _closed_ fold can be marked.
if line('w0') > marknum
while line('w0') > marknum
exe "norm! \<C-y>"
endwhile
if line('w0') != marknum
exe "norm! \<C-e>H"
endif
# Handle non-wrapped lines.
elseif line('w0') < marknum
while line('w0') < marknum
exe "norm! \<C-e>"
endwhile
if line('w0') != marknum
exe "norm! \<C-y>H"
endif
endif
norm! ^
return AssertCursorForwardProgress()
enddef
END
let MAX_FAILED_COUNT = 5
let DUMP_OPTS = exists("$VIM_SYNTAX_TEST_WAIT_TIME") &&
\ !empty($VIM_SYNTAX_TEST_WAIT_TIME)
\ ? {'wait': max([1, str2nr($VIM_SYNTAX_TEST_WAIT_TIME)])}
\ : {}
lockvar DUMP_OPTS MAX_FAILED_COUNT XTESTSCRIPT
let ok_count = 0
runtime(syntax-tests): Do not ignore failed screendumps The process of preparing and submitting syntax tests is fraught with challenges that can turn away many aspiring contributors from ever attempting it. (Out of 69 languages introduced since v9.0.1627, there are only syntax tests for Tera.) After v9.1.1176~1, one visual clue for admitting syntax test failures previously available with e.g. "git status" is gone after all files under "failed/" have been made ignored for Git and Mercurial. There isn't a single way to go about it: some people may move files from "failed/" to "dumps/" after each iteration; some people may only move "good" iteration files; when a test file is refactored to a great extent, some people may prefer deleting all test-related files under "dumps/" before moving files from "failed/". The usability of reporting, at any time, that there are some _untracked_ files under "failed/" cannot be overstated. Without it, the chances are greater for pushing mismatched changesets. And when tests fail then everyone but the author will be kept in the dark about the cause: were some updated screendumps not committed _or_ was a wrong version of the syntax plugin committed? Another file, "testdir/Xfilter" (v9.1.0763), that will be created to establish communication from Make to Vim about what subset of syntax tests is requested for running, should also be not ignored but rather deleted once its contents are read. Unless it is explicitly deleted _after test failure_, the file may contain new *and* old test names when another testing attempt is under way. And by virtue of it being ignored, the reason for also running not requested tests will be as ever puzzling. Both Git and Mercurial support per-user configuration; such wide-reaching settings hardly belong to clonable defaults. Also, match literal dots in testname filters. Also, discover and report _some_ disused screendump files tracked under "dumps/". References: - https://git-scm.com/docs/gitignore - https://www.mercurial-scm.org/help/topics/config#ui closes: #16917 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2025-03-23 10:42:23 +01:00
let disused_pages = []
let failed_tests = []
let skipped_count = 0
let last_test_status = 'invalid'
runtime(syntax-tests): Do not ignore failed screendumps The process of preparing and submitting syntax tests is fraught with challenges that can turn away many aspiring contributors from ever attempting it. (Out of 69 languages introduced since v9.0.1627, there are only syntax tests for Tera.) After v9.1.1176~1, one visual clue for admitting syntax test failures previously available with e.g. "git status" is gone after all files under "failed/" have been made ignored for Git and Mercurial. There isn't a single way to go about it: some people may move files from "failed/" to "dumps/" after each iteration; some people may only move "good" iteration files; when a test file is refactored to a great extent, some people may prefer deleting all test-related files under "dumps/" before moving files from "failed/". The usability of reporting, at any time, that there are some _untracked_ files under "failed/" cannot be overstated. Without it, the chances are greater for pushing mismatched changesets. And when tests fail then everyone but the author will be kept in the dark about the cause: were some updated screendumps not committed _or_ was a wrong version of the syntax plugin committed? Another file, "testdir/Xfilter" (v9.1.0763), that will be created to establish communication from Make to Vim about what subset of syntax tests is requested for running, should also be not ignored but rather deleted once its contents are read. Unless it is explicitly deleted _after test failure_, the file may contain new *and* old test names when another testing attempt is under way. And by virtue of it being ignored, the reason for also running not requested tests will be as ever puzzling. Both Git and Mercurial support per-user configuration; such wide-reaching settings hardly belong to clonable defaults. Also, match literal dots in testname filters. Also, discover and report _some_ disused screendump files tracked under "dumps/". References: - https://git-scm.com/docs/gitignore - https://www.mercurial-scm.org/help/topics/config#ui closes: #16917 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2025-03-23 10:42:23 +01:00
let filter = ''
" Create a map of setup configuration filenames with their basenames as keys.
let setup = glob('input/setup/*.vim', 1, 1)
\ ->reduce({d, f -> extend(d, {fnamemodify(f, ':t:r'): f})}, {})
" Turn a subset of filenames etc. requested for testing into a pattern.
runtime(syntax-tests): Do not ignore failed screendumps The process of preparing and submitting syntax tests is fraught with challenges that can turn away many aspiring contributors from ever attempting it. (Out of 69 languages introduced since v9.0.1627, there are only syntax tests for Tera.) After v9.1.1176~1, one visual clue for admitting syntax test failures previously available with e.g. "git status" is gone after all files under "failed/" have been made ignored for Git and Mercurial. There isn't a single way to go about it: some people may move files from "failed/" to "dumps/" after each iteration; some people may only move "good" iteration files; when a test file is refactored to a great extent, some people may prefer deleting all test-related files under "dumps/" before moving files from "failed/". The usability of reporting, at any time, that there are some _untracked_ files under "failed/" cannot be overstated. Without it, the chances are greater for pushing mismatched changesets. And when tests fail then everyone but the author will be kept in the dark about the cause: were some updated screendumps not committed _or_ was a wrong version of the syntax plugin committed? Another file, "testdir/Xfilter" (v9.1.0763), that will be created to establish communication from Make to Vim about what subset of syntax tests is requested for running, should also be not ignored but rather deleted once its contents are read. Unless it is explicitly deleted _after test failure_, the file may contain new *and* old test names when another testing attempt is under way. And by virtue of it being ignored, the reason for also running not requested tests will be as ever puzzling. Both Git and Mercurial support per-user configuration; such wide-reaching settings hardly belong to clonable defaults. Also, match literal dots in testname filters. Also, discover and report _some_ disused screendump files tracked under "dumps/". References: - https://git-scm.com/docs/gitignore - https://www.mercurial-scm.org/help/topics/config#ui closes: #16917 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2025-03-23 10:42:23 +01:00
if filereadable('../testdir/Xfilter')
let filter = readfile('../testdir/Xfilter')
\ ->map({_, v -> '^' .. escape(substitute(v, '_$', '', ''), '.')})
\ ->join('\|')
runtime(syntax-tests): Do not ignore failed screendumps The process of preparing and submitting syntax tests is fraught with challenges that can turn away many aspiring contributors from ever attempting it. (Out of 69 languages introduced since v9.0.1627, there are only syntax tests for Tera.) After v9.1.1176~1, one visual clue for admitting syntax test failures previously available with e.g. "git status" is gone after all files under "failed/" have been made ignored for Git and Mercurial. There isn't a single way to go about it: some people may move files from "failed/" to "dumps/" after each iteration; some people may only move "good" iteration files; when a test file is refactored to a great extent, some people may prefer deleting all test-related files under "dumps/" before moving files from "failed/". The usability of reporting, at any time, that there are some _untracked_ files under "failed/" cannot be overstated. Without it, the chances are greater for pushing mismatched changesets. And when tests fail then everyone but the author will be kept in the dark about the cause: were some updated screendumps not committed _or_ was a wrong version of the syntax plugin committed? Another file, "testdir/Xfilter" (v9.1.0763), that will be created to establish communication from Make to Vim about what subset of syntax tests is requested for running, should also be not ignored but rather deleted once its contents are read. Unless it is explicitly deleted _after test failure_, the file may contain new *and* old test names when another testing attempt is under way. And by virtue of it being ignored, the reason for also running not requested tests will be as ever puzzling. Both Git and Mercurial support per-user configuration; such wide-reaching settings hardly belong to clonable defaults. Also, match literal dots in testname filters. Also, discover and report _some_ disused screendump files tracked under "dumps/". References: - https://git-scm.com/docs/gitignore - https://www.mercurial-scm.org/help/topics/config#ui closes: #16917 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2025-03-23 10:42:23 +01:00
call delete('../testdir/Xfilter')
endif
" Treat "^self-testing" as a string NOT as a regexp.
if filter ==# '^self-testing'
let dirpath = 'input/selftestdir/'
runtime(syntax-tests): Facilitate the viewing of rendered screendumps With the submitted "viewdumps.vim" script, a few manual steps in typical workflows (see below) can be automated. The updated "README.txt" contains additional information. ============================================================ Reviewing LOCAL failed syntax tests can be arranged as follows: 1) Run tests and generate screendumps: ------------------------------------------------------------ cd /path/to/fork/runtime/syntax make clean test ------------------------------------------------------------ 2) Examine the screendumps from the "failed" directory: ------------------------------------------------------------ ../../src/vim --clean -S testdir/viewdumps.vim ------------------------------------------------------------ ============================================================ Reviewing UPLOADED failed syntax tests can be arranged as follows (it can be further locally scripted): 1) Fetch an artifact with failed screendumps from "github.com/vim/vim/actions/runs/A_ID/artifacts/B_ID". 2) Extract the archived files: ------------------------------------------------------------ unzip /tmp/failed-tests.zip -d /tmp ------------------------------------------------------------ 3) Set up the "dumps" directory. Create a symlink to "/path/to/fork/dirs/dumps" in the extracted directories so that term_dumpdiff() can be used. (The lookup algorithm resolves "dumps" for every loaded filename. So, with "/tmp/runtime/syntax/testdir/failed/*.dump" files passed as script arguments, the algorithm will make the files in "/tmp/runtime/syntax/testdir/dumps" queried.) ------------------------------------------------------------ cd /path/to/fork ln -s $(pwd)/runtime/syntax/testdir/dumps \ /tmp/runtime/syntax/testdir/dumps ------------------------------------------------------------ 4) Examine the extracted screendumps: ------------------------------------------------------------ ./src/vim --clean -S runtime/syntax/testdir/viewdumps.vim \ /tmp/runtime/syntax/testdir/failed/*.dump ------------------------------------------------------------ 5) Clean up: ------------------------------------------------------------ unlink /tmp/runtime/syntax/testdir/dumps rm -rf /tmp/runtime ------------------------------------------------------------ ============================================================ Reviewing SUBMITTED FOR PULL REQUEST syntax tests can be arranged as follows (it can be further locally scripted): 1) List the fetched changeset and write the changed "dumps" filenames to "/tmp/filelist": ------------------------------------------------------------ cd /path/to/fork git switch prs/1234 git diff-index --relative=runtime/syntax/testdir/dumps/ \ --name-only prs/1234~1 > /tmp/filelist ------------------------------------------------------------ 2) Reconcile relative filepaths, and copy next-to-be-updated "dumps" files in the "failed" directory (note the missing new screendumps, if any): ------------------------------------------------------------ git switch master cd runtime/syntax/testdir/dumps cp -t ../failed $(cat /tmp/filelist) ------------------------------------------------------------ 3) Remember about the introduced INVERTED relation between "dumps" and "failed", i.e. the files to be committed are in "dumps" already and their previous versions are in "failed"; therefore, copy the missing new screendumps from "dumps" to "failed" (otherwise these won't be shown): ------------------------------------------------------------ git switch prs/1234 cp -t ../failed foo_10.dump foo_11.dump foo_12.dump ------------------------------------------------------------ 4) Examine the screendumps from the "failed" directory (new screendumps will be shown with no difference between their versions): ------------------------------------------------------------ cd .. ../../../src/vim --clean -S viewdumps.vim ------------------------------------------------------------ closes: #15476 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2024-08-12 18:37:15 +02:00
let fnames = readdir(dirpath, {fname -> fname !~ '^README\.txt$'})
else
let dirpath = 'input/'
let filter ..= exists("$VIM_SYNTAX_TEST_FILTER") &&
\ !empty($VIM_SYNTAX_TEST_FILTER)
\ ? (empty(filter) ? '' : '\|') .. $VIM_SYNTAX_TEST_FILTER
\ : ''
let fnames = readdir(dirpath,
\ {subset -> {fname -> fname !~ '\~$' && fname =~# subset}}(
\ empty(filter) ? '^.\+\..\+$' : filter))
endif
for fname in fnames
let root = fnamemodify(fname, ':r')
let fname = dirpath .. fname
" Execute the test if the "done" file does not exist or when the input file
" is newer.
let in_time = getftime(fname)
let out_time = getftime('done/' .. root)
if out_time < 0 || in_time > out_time
call ch_log('running tests for: ' .. fname)
let filetype = substitute(root, '\([^_.]*\)[_.].*', '\1', '')
let failed_root = 'failed/' .. root
runtime(syntax-tests): Do not ignore failed screendumps The process of preparing and submitting syntax tests is fraught with challenges that can turn away many aspiring contributors from ever attempting it. (Out of 69 languages introduced since v9.0.1627, there are only syntax tests for Tera.) After v9.1.1176~1, one visual clue for admitting syntax test failures previously available with e.g. "git status" is gone after all files under "failed/" have been made ignored for Git and Mercurial. There isn't a single way to go about it: some people may move files from "failed/" to "dumps/" after each iteration; some people may only move "good" iteration files; when a test file is refactored to a great extent, some people may prefer deleting all test-related files under "dumps/" before moving files from "failed/". The usability of reporting, at any time, that there are some _untracked_ files under "failed/" cannot be overstated. Without it, the chances are greater for pushing mismatched changesets. And when tests fail then everyone but the author will be kept in the dark about the cause: were some updated screendumps not committed _or_ was a wrong version of the syntax plugin committed? Another file, "testdir/Xfilter" (v9.1.0763), that will be created to establish communication from Make to Vim about what subset of syntax tests is requested for running, should also be not ignored but rather deleted once its contents are read. Unless it is explicitly deleted _after test failure_, the file may contain new *and* old test names when another testing attempt is under way. And by virtue of it being ignored, the reason for also running not requested tests will be as ever puzzling. Both Git and Mercurial support per-user configuration; such wide-reaching settings hardly belong to clonable defaults. Also, match literal dots in testname filters. Also, discover and report _some_ disused screendump files tracked under "dumps/". References: - https://git-scm.com/docs/gitignore - https://www.mercurial-scm.org/help/topics/config#ui closes: #16917 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2025-03-23 10:42:23 +01:00
for pagename in glob(failed_root .. '_\d*\.dump', 1, 1)
call delete(pagename)
endfor
call delete('done/' .. root)
call writefile(XTESTSCRIPT, 'Xtestscript')
" close all but the last window
while winnr('$') > 1
close
endwhile
" Redraw to make sure that messages are cleared and there is enough space
" for the terminal window.
redraw
" Let "Xtestscript#SetUpVim()" turn the syntax on.
let prefix = '-Nu NONE -S Xtestscript'
let path = get(setup, root, '')
" Source the found setup configuration file.
let args = !empty(path)
\ ? prefix .. ' -S ' .. path
\ : prefix
let buf = RunVimInTerminal(args, {})
" edit the file only after catching the SwapExists event
call term_sendkeys(buf, ":edit " .. fname .. "\<CR>")
" set up the testing environment
call term_sendkeys(buf, ":call SetUpVim()\<CR>")
" load filetype specific settings
call term_sendkeys(buf, ":call LoadFiletype('" .. filetype .. "')\<CR>")
" Make a synchronisation point between buffers by requesting to echo
" a known token in the terminal buffer and asserting its availability
" with "s:TermWaitAndPollRuler()".
if filetype == 'sh'
call term_sendkeys(buf, ":call ShellInfo()\<CR>")
else
call term_sendkeys(buf, ":echo 'is_nonce'\<CR>")
endif
let root_00 = root .. '_00'
runtime(syntax-tests): Allow for folded and wrapped lines in syntax test files The current implementation falls short for syntax test files on two accounts: 1. With folded lines -- some lines before folded lines are unnecessarily repeated in generated dump files because closed folded lines are always treated as opened for the cursor to move _in_ instead of to move _over_ them. 2. With wrapped lines (longer than 75 columns) -- some lines are omitted in generated dump files because calculations for the cursor progress and its movement commands only refer to file lines and not their layout within a 20x75 buffer (less &cmdheight). As an alternative, we abandon deterministic (and inaccurate at times) calculations for the cursor progress and, instead, advance the cursor by as much as before for a single dump file, but now rely on marking the last visible line and additional movement to position lines at desired offsets, carefully preserving compatibility for the &scrolloff and &ruler values inherited from defaults.vim. The parent Vim process will keep track of progress through a syntax test file made by its child process ("terminal") by reading the rightmost end of the ruler line from the terminal buffer, looking for " All " or " Bot " for its cue to finish dump file generation. With these changes applied, the lossless line length limit will be raised from 75 to 1425 (for a 19x75 view) columns. Also, prefer "lastline" to "truncate" for &display; hiding the content of any last _long_ line in a view goes against the purpose of syntax file testing -- all lines should be recorded. related: #15150 fixes: #14245 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2024-05-21 01:10:26 +03:00
let in_name_and_out_name = fname .. ': failed/' .. root_00 .. '.dump'
" Queue up all "term_sendkeys()"es and let them finish before returning
" from "s:TermWaitAndPollRuler()".
let ruler = s:TermWaitAndPollRuler(buf, in_name_and_out_name)
runtime(syntax-tests): Allow for folded and wrapped lines in syntax test files The current implementation falls short for syntax test files on two accounts: 1. With folded lines -- some lines before folded lines are unnecessarily repeated in generated dump files because closed folded lines are always treated as opened for the cursor to move _in_ instead of to move _over_ them. 2. With wrapped lines (longer than 75 columns) -- some lines are omitted in generated dump files because calculations for the cursor progress and its movement commands only refer to file lines and not their layout within a 20x75 buffer (less &cmdheight). As an alternative, we abandon deterministic (and inaccurate at times) calculations for the cursor progress and, instead, advance the cursor by as much as before for a single dump file, but now rely on marking the last visible line and additional movement to position lines at desired offsets, carefully preserving compatibility for the &scrolloff and &ruler values inherited from defaults.vim. The parent Vim process will keep track of progress through a syntax test file made by its child process ("terminal") by reading the rightmost end of the ruler line from the terminal buffer, looking for " All " or " Bot " for its cue to finish dump file generation. With these changes applied, the lossless line length limit will be raised from 75 to 1425 (for a 19x75 view) columns. Also, prefer "lastline" to "truncate" for &display; hiding the content of any last _long_ line in a view goes against the purpose of syntax file testing -- all lines should be recorded. related: #15150 fixes: #14245 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2024-05-21 01:10:26 +03:00
call ch_log('First screendump for ' .. in_name_and_out_name)
" Make a screendump at the start of the file: failed/root_00.dump
let fail = VerifyScreenDump(buf, root_00, DUMP_OPTS)
runtime(syntax-tests): Do not ignore failed screendumps The process of preparing and submitting syntax tests is fraught with challenges that can turn away many aspiring contributors from ever attempting it. (Out of 69 languages introduced since v9.0.1627, there are only syntax tests for Tera.) After v9.1.1176~1, one visual clue for admitting syntax test failures previously available with e.g. "git status" is gone after all files under "failed/" have been made ignored for Git and Mercurial. There isn't a single way to go about it: some people may move files from "failed/" to "dumps/" after each iteration; some people may only move "good" iteration files; when a test file is refactored to a great extent, some people may prefer deleting all test-related files under "dumps/" before moving files from "failed/". The usability of reporting, at any time, that there are some _untracked_ files under "failed/" cannot be overstated. Without it, the chances are greater for pushing mismatched changesets. And when tests fail then everyone but the author will be kept in the dark about the cause: were some updated screendumps not committed _or_ was a wrong version of the syntax plugin committed? Another file, "testdir/Xfilter" (v9.1.0763), that will be created to establish communication from Make to Vim about what subset of syntax tests is requested for running, should also be not ignored but rather deleted once its contents are read. Unless it is explicitly deleted _after test failure_, the file may contain new *and* old test names when another testing attempt is under way. And by virtue of it being ignored, the reason for also running not requested tests will be as ever puzzling. Both Git and Mercurial support per-user configuration; such wide-reaching settings hardly belong to clonable defaults. Also, match literal dots in testname filters. Also, discover and report _some_ disused screendump files tracked under "dumps/". References: - https://git-scm.com/docs/gitignore - https://www.mercurial-scm.org/help/topics/config#ui closes: #16917 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2025-03-23 10:42:23 +01:00
let nr = 0
" Accommodate the next code block to "buf"'s contingency for self
" wipe-out.
try
let keys_a = ":call ScrollToSecondPage((18 * 75 + 1), 19, 5) | redraw!\<CR>"
let keys_b = ":call ScrollToNextPage((18 * 75 + 1), 19, 5) | redraw!\<CR>"
while s:CannotSeeLastLine(ruler)
call term_sendkeys(buf, keys_a)
let keys_a = keys_b
runtime(syntax-tests): Allow for folded and wrapped lines in syntax test files The current implementation falls short for syntax test files on two accounts: 1. With folded lines -- some lines before folded lines are unnecessarily repeated in generated dump files because closed folded lines are always treated as opened for the cursor to move _in_ instead of to move _over_ them. 2. With wrapped lines (longer than 75 columns) -- some lines are omitted in generated dump files because calculations for the cursor progress and its movement commands only refer to file lines and not their layout within a 20x75 buffer (less &cmdheight). As an alternative, we abandon deterministic (and inaccurate at times) calculations for the cursor progress and, instead, advance the cursor by as much as before for a single dump file, but now rely on marking the last visible line and additional movement to position lines at desired offsets, carefully preserving compatibility for the &scrolloff and &ruler values inherited from defaults.vim. The parent Vim process will keep track of progress through a syntax test file made by its child process ("terminal") by reading the rightmost end of the ruler line from the terminal buffer, looking for " All " or " Bot " for its cue to finish dump file generation. With these changes applied, the lossless line length limit will be raised from 75 to 1425 (for a 19x75 view) columns. Also, prefer "lastline" to "truncate" for &display; hiding the content of any last _long_ line in a view goes against the purpose of syntax file testing -- all lines should be recorded. related: #15150 fixes: #14245 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2024-05-21 01:10:26 +03:00
let nr += 1
let root_next = printf('%s_%02d', root, nr)
let in_name_and_out_name = fname .. ': failed/' .. root_next .. '.dump'
let ruler = s:TermPollRuler(
\ function('s:CannotDumpNextPage', [buf, ruler]),
\ buf,
\ in_name_and_out_name)
call ch_log('Next screendump for ' .. in_name_and_out_name)
" Make a screendump of every 18 lines of the file: failed/root_NN.dump
let fail += VerifyScreenDump(buf, root_next, DUMP_OPTS)
endwhile
call StopVimInTerminal(buf)
finally
call delete('Xtestscript')
endtry
runtime(syntax-tests): Do not ignore failed screendumps The process of preparing and submitting syntax tests is fraught with challenges that can turn away many aspiring contributors from ever attempting it. (Out of 69 languages introduced since v9.0.1627, there are only syntax tests for Tera.) After v9.1.1176~1, one visual clue for admitting syntax test failures previously available with e.g. "git status" is gone after all files under "failed/" have been made ignored for Git and Mercurial. There isn't a single way to go about it: some people may move files from "failed/" to "dumps/" after each iteration; some people may only move "good" iteration files; when a test file is refactored to a great extent, some people may prefer deleting all test-related files under "dumps/" before moving files from "failed/". The usability of reporting, at any time, that there are some _untracked_ files under "failed/" cannot be overstated. Without it, the chances are greater for pushing mismatched changesets. And when tests fail then everyone but the author will be kept in the dark about the cause: were some updated screendumps not committed _or_ was a wrong version of the syntax plugin committed? Another file, "testdir/Xfilter" (v9.1.0763), that will be created to establish communication from Make to Vim about what subset of syntax tests is requested for running, should also be not ignored but rather deleted once its contents are read. Unless it is explicitly deleted _after test failure_, the file may contain new *and* old test names when another testing attempt is under way. And by virtue of it being ignored, the reason for also running not requested tests will be as ever puzzling. Both Git and Mercurial support per-user configuration; such wide-reaching settings hardly belong to clonable defaults. Also, match literal dots in testname filters. Also, discover and report _some_ disused screendump files tracked under "dumps/". References: - https://git-scm.com/docs/gitignore - https://www.mercurial-scm.org/help/topics/config#ui closes: #16917 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2025-03-23 10:42:23 +01:00
let nr += 1
let pagename = printf('dumps/%s_%02d.dump', root, nr)
while filereadable(pagename)
call add(disused_pages, pagename)
let nr += 1
let pagename = printf('dumps/%s_%02d.dump', root, nr)
endwhile
" redraw here to avoid the following messages to get mixed up with screen
" output.
redraw
" Add any assert errors to s:messages.
if len(v:errors) > 0
call extend(s:messages, v:errors)
if last_test_status == 'passed'
call EraseLineAndReturnCarriage('Test ' .. root .. ' OK')
else
echon "\n"
endif
" Echo the errors here, in case the script aborts or the "messages" file
" is not displayed later.
echomsg v:errors
let v:errors = []
let fail += 1
endif
if fail == 0
if last_test_status == 'skipped'
echon "\n"
endif
let last_test_status = 'passed'
let msg = "Test " .. root .. " OK"
call Message(msg)
call EraseLineAndReturnCarriage(msg)
call writefile(['OK'], 'done/' .. root)
let ok_count += 1
else
let last_test_status = 'failed'
call Message("Test " .. root .. " FAILED")
echon "\n"
call delete('done/' .. root)
eval failed_tests->add(root)
if len(failed_tests) > MAX_FAILED_COUNT
call Message('')
call Message('Too many errors, aborting')
endif
endif
else
if last_test_status == 'passed'
call EraseLineAndReturnCarriage('Test ' .. root .. ' OK')
endif
let last_test_status = 'skipped'
call Message("Test " .. root .. " skipped")
let skipped_count += 1
endif
" Append messages to the file "testdir/messages"
call AppendMessages('Input file ' .. fname .. ':')
if len(failed_tests) > MAX_FAILED_COUNT
break
endif
endfor
if last_test_status == 'passed' && exists('root')
call EraseLineAndReturnCarriage('Test ' .. root .. ' OK')
endif
call Message(s:test_run_message)
call Message('OK: ' .. ok_count)
call Message('FAILED: ' .. len(failed_tests) .. ': ' .. string(failed_tests))
call Message('skipped: ' .. skipped_count)
runtime(syntax-tests): Facilitate the viewing of rendered screendumps With the submitted "viewdumps.vim" script, a few manual steps in typical workflows (see below) can be automated. The updated "README.txt" contains additional information. ============================================================ Reviewing LOCAL failed syntax tests can be arranged as follows: 1) Run tests and generate screendumps: ------------------------------------------------------------ cd /path/to/fork/runtime/syntax make clean test ------------------------------------------------------------ 2) Examine the screendumps from the "failed" directory: ------------------------------------------------------------ ../../src/vim --clean -S testdir/viewdumps.vim ------------------------------------------------------------ ============================================================ Reviewing UPLOADED failed syntax tests can be arranged as follows (it can be further locally scripted): 1) Fetch an artifact with failed screendumps from "github.com/vim/vim/actions/runs/A_ID/artifacts/B_ID". 2) Extract the archived files: ------------------------------------------------------------ unzip /tmp/failed-tests.zip -d /tmp ------------------------------------------------------------ 3) Set up the "dumps" directory. Create a symlink to "/path/to/fork/dirs/dumps" in the extracted directories so that term_dumpdiff() can be used. (The lookup algorithm resolves "dumps" for every loaded filename. So, with "/tmp/runtime/syntax/testdir/failed/*.dump" files passed as script arguments, the algorithm will make the files in "/tmp/runtime/syntax/testdir/dumps" queried.) ------------------------------------------------------------ cd /path/to/fork ln -s $(pwd)/runtime/syntax/testdir/dumps \ /tmp/runtime/syntax/testdir/dumps ------------------------------------------------------------ 4) Examine the extracted screendumps: ------------------------------------------------------------ ./src/vim --clean -S runtime/syntax/testdir/viewdumps.vim \ /tmp/runtime/syntax/testdir/failed/*.dump ------------------------------------------------------------ 5) Clean up: ------------------------------------------------------------ unlink /tmp/runtime/syntax/testdir/dumps rm -rf /tmp/runtime ------------------------------------------------------------ ============================================================ Reviewing SUBMITTED FOR PULL REQUEST syntax tests can be arranged as follows (it can be further locally scripted): 1) List the fetched changeset and write the changed "dumps" filenames to "/tmp/filelist": ------------------------------------------------------------ cd /path/to/fork git switch prs/1234 git diff-index --relative=runtime/syntax/testdir/dumps/ \ --name-only prs/1234~1 > /tmp/filelist ------------------------------------------------------------ 2) Reconcile relative filepaths, and copy next-to-be-updated "dumps" files in the "failed" directory (note the missing new screendumps, if any): ------------------------------------------------------------ git switch master cd runtime/syntax/testdir/dumps cp -t ../failed $(cat /tmp/filelist) ------------------------------------------------------------ 3) Remember about the introduced INVERTED relation between "dumps" and "failed", i.e. the files to be committed are in "dumps" already and their previous versions are in "failed"; therefore, copy the missing new screendumps from "dumps" to "failed" (otherwise these won't be shown): ------------------------------------------------------------ git switch prs/1234 cp -t ../failed foo_10.dump foo_11.dump foo_12.dump ------------------------------------------------------------ 4) Examine the screendumps from the "failed" directory (new screendumps will be shown with no difference between their versions): ------------------------------------------------------------ cd .. ../../../src/vim --clean -S viewdumps.vim ------------------------------------------------------------ closes: #15476 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2024-08-12 18:37:15 +02:00
runtime(syntax-tests): Do not ignore failed screendumps The process of preparing and submitting syntax tests is fraught with challenges that can turn away many aspiring contributors from ever attempting it. (Out of 69 languages introduced since v9.0.1627, there are only syntax tests for Tera.) After v9.1.1176~1, one visual clue for admitting syntax test failures previously available with e.g. "git status" is gone after all files under "failed/" have been made ignored for Git and Mercurial. There isn't a single way to go about it: some people may move files from "failed/" to "dumps/" after each iteration; some people may only move "good" iteration files; when a test file is refactored to a great extent, some people may prefer deleting all test-related files under "dumps/" before moving files from "failed/". The usability of reporting, at any time, that there are some _untracked_ files under "failed/" cannot be overstated. Without it, the chances are greater for pushing mismatched changesets. And when tests fail then everyone but the author will be kept in the dark about the cause: were some updated screendumps not committed _or_ was a wrong version of the syntax plugin committed? Another file, "testdir/Xfilter" (v9.1.0763), that will be created to establish communication from Make to Vim about what subset of syntax tests is requested for running, should also be not ignored but rather deleted once its contents are read. Unless it is explicitly deleted _after test failure_, the file may contain new *and* old test names when another testing attempt is under way. And by virtue of it being ignored, the reason for also running not requested tests will be as ever puzzling. Both Git and Mercurial support per-user configuration; such wide-reaching settings hardly belong to clonable defaults. Also, match literal dots in testname filters. Also, discover and report _some_ disused screendump files tracked under "dumps/". References: - https://git-scm.com/docs/gitignore - https://www.mercurial-scm.org/help/topics/config#ui closes: #16917 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2025-03-23 10:42:23 +01:00
for pagename in disused_pages
call Message(printf('No input page found for "%s"', pagename))
endfor
runtime(syntax-tests): Facilitate the viewing of rendered screendumps With the submitted "viewdumps.vim" script, a few manual steps in typical workflows (see below) can be automated. The updated "README.txt" contains additional information. ============================================================ Reviewing LOCAL failed syntax tests can be arranged as follows: 1) Run tests and generate screendumps: ------------------------------------------------------------ cd /path/to/fork/runtime/syntax make clean test ------------------------------------------------------------ 2) Examine the screendumps from the "failed" directory: ------------------------------------------------------------ ../../src/vim --clean -S testdir/viewdumps.vim ------------------------------------------------------------ ============================================================ Reviewing UPLOADED failed syntax tests can be arranged as follows (it can be further locally scripted): 1) Fetch an artifact with failed screendumps from "github.com/vim/vim/actions/runs/A_ID/artifacts/B_ID". 2) Extract the archived files: ------------------------------------------------------------ unzip /tmp/failed-tests.zip -d /tmp ------------------------------------------------------------ 3) Set up the "dumps" directory. Create a symlink to "/path/to/fork/dirs/dumps" in the extracted directories so that term_dumpdiff() can be used. (The lookup algorithm resolves "dumps" for every loaded filename. So, with "/tmp/runtime/syntax/testdir/failed/*.dump" files passed as script arguments, the algorithm will make the files in "/tmp/runtime/syntax/testdir/dumps" queried.) ------------------------------------------------------------ cd /path/to/fork ln -s $(pwd)/runtime/syntax/testdir/dumps \ /tmp/runtime/syntax/testdir/dumps ------------------------------------------------------------ 4) Examine the extracted screendumps: ------------------------------------------------------------ ./src/vim --clean -S runtime/syntax/testdir/viewdumps.vim \ /tmp/runtime/syntax/testdir/failed/*.dump ------------------------------------------------------------ 5) Clean up: ------------------------------------------------------------ unlink /tmp/runtime/syntax/testdir/dumps rm -rf /tmp/runtime ------------------------------------------------------------ ============================================================ Reviewing SUBMITTED FOR PULL REQUEST syntax tests can be arranged as follows (it can be further locally scripted): 1) List the fetched changeset and write the changed "dumps" filenames to "/tmp/filelist": ------------------------------------------------------------ cd /path/to/fork git switch prs/1234 git diff-index --relative=runtime/syntax/testdir/dumps/ \ --name-only prs/1234~1 > /tmp/filelist ------------------------------------------------------------ 2) Reconcile relative filepaths, and copy next-to-be-updated "dumps" files in the "failed" directory (note the missing new screendumps, if any): ------------------------------------------------------------ git switch master cd runtime/syntax/testdir/dumps cp -t ../failed $(cat /tmp/filelist) ------------------------------------------------------------ 3) Remember about the introduced INVERTED relation between "dumps" and "failed", i.e. the files to be committed are in "dumps" already and their previous versions are in "failed"; therefore, copy the missing new screendumps from "dumps" to "failed" (otherwise these won't be shown): ------------------------------------------------------------ git switch prs/1234 cp -t ../failed foo_10.dump foo_11.dump foo_12.dump ------------------------------------------------------------ 4) Examine the screendumps from the "failed" directory (new screendumps will be shown with no difference between their versions): ------------------------------------------------------------ cd .. ../../../src/vim --clean -S viewdumps.vim ------------------------------------------------------------ closes: #15476 Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
2024-08-12 18:37:15 +02:00
if !empty(failed_tests)
call Message('')
call Message('View generated screendumps with "../../src/vim --clean -S testdir/viewdumps.vim"')
endif
call AppendMessages('== SUMMARY SYNTAX TESTS ==')
if len(failed_tests) > 0
" have make report an error
cquit
endif
endfunc
call RunTest()
" Matching "if 1" at the start.
endif
qall!
" vim:sw=2:ts=8:noet: