From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from [87.239.111.99] (localhost [127.0.0.1]) by dev.tarantool.org (Postfix) with ESMTP id BCF636EC56; Tue, 16 Mar 2021 17:51:11 +0300 (MSK) DKIM-Filter: OpenDKIM Filter v2.11.0 dev.tarantool.org BCF636EC56 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=tarantool.org; s=dev; t=1615906271; bh=Kc8hooaNgrcO74SZ0F2lWs+S/WzkmFLaGz+Rls9gLw8=; h=Date:In-Reply-To:To:References:Subject:List-Id:List-Unsubscribe: List-Archive:List-Post:List-Help:List-Subscribe:From:Reply-To:Cc: From; b=njpqw27MsZcoaOiN0X+JKeBsntnmyKp26IqYOYv/DrjzMhjXtpwYp16OtQoME4G7C c4xgFpDSKJg7ZBKWfKcBti2NMNyDE+XAQgTn0LO6VTZzdCCiQF5WHALEssxIgyBMkG UE2QlbkaZHN5/H6Dz/ihOXU7r1xqKgr3At2U6I/w= Received: from smtp42.i.mail.ru (smtp42.i.mail.ru [94.100.177.102]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by dev.tarantool.org (Postfix) with ESMTPS id 941B26EC56 for ; Tue, 16 Mar 2021 17:51:09 +0300 (MSK) DKIM-Filter: OpenDKIM Filter v2.11.0 dev.tarantool.org 941B26EC56 Received: by smtp42.i.mail.ru with esmtpa (envelope-from ) id 1lMB2K-0007nZ-HK; Tue, 16 Mar 2021 17:51:09 +0300 Message-Id: Content-Type: multipart/alternative; boundary="Apple-Mail=_5EE338FA-C215-4152-AD49-8B967C912BAD" Mime-Version: 1.0 (Mac OS X Mail 14.0 \(3654.60.0.2.21\)) Date: Tue, 16 Mar 2021 17:51:08 +0300 In-Reply-To: <20210316105142.GK9042@tarantool.org> To: Igor Munkin References: <31d2b8f880693c4af3d47d7c6693acfd79459052.1615819534.git.skaplun@tarantool.org> <20210315174421.GG9042@tarantool.org> <20210316060142.GD16737@root> <20210316105142.GK9042@tarantool.org> X-Mailer: Apple Mail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eAau8CL7WIMRKs4sN3D3tLDjz0dLbV79QFUyzQ2Ujvy7cMT6pYYqY16iZVKkSc3dCLJ7zSJH7+u4VD18S7Vl4ZUrpaVfd2+vE6kuoey4m4VkSEu530nj6fImhcD4MUrOEAnl0W826KZ9Q+tr5ycPtXkTV4k65bRjmOUUP8cvGozZ33TWg5HZplvhhXbhDGzqmQDTd6OAevLeAnq3Ra9uf7zvY2zzsIhlcp/Y7m53TZgf2aB4JOg4gkr2biojwV/GgY7Z4vUVHeFLahKP5w== X-Mailru-Sender: 3B9A0136629DC912F4AABCEFC589C81E82887788639C5A4253D40B7EC5C4AE76F5494DF53083D48CAD07DD1419AC565FA614486B47F28B67C5E079CCF3B0523AED31B7EB2E253A9E112434F685709FCF0DA7A0AF5A3A8387 X-Mras: Ok Subject: Re: [Tarantool-patches] [PATCH v2 luajit 3/5] test: adjust lua-Harness test suite for Tarantool X-BeenThere: tarantool-patches@dev.tarantool.org X-Mailman-Version: 2.1.34 Precedence: list List-Id: Tarantool development patches List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , From: Sergey Ostanevich via Tarantool-patches Reply-To: Sergey Ostanevich Cc: tarantool-patches@dev.tarantool.org Errors-To: tarantool-patches-bounces@dev.tarantool.org Sender: "Tarantool-patches" --Apple-Mail=_5EE338FA-C215-4152-AD49-8B967C912BAD Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=utf-8 Hi! Thanks for patch and review! With all comments below fixed the patch is LGTM. My proposal on dofile_fullpath() naming: - split the functions, it=E2=80=99ll ease reading. Keep dofile() in the = place of use,=20 wrap the name with additional function - pick a name that echo what function does: delivers a filename relative = to the test itself e.g. relative_to_test() This results in dofile(relative_to_test('lexico53/boolean.t=E2=80=99)) regards, Sergos > On 16 Mar 2021, at 13:51, Igor Munkin wrote: >=20 > Sergey, > =20 > On 16.03.21, Sergey Kaplun wrote: >> Igor, >>=20 >> Thanks for the review! >>=20 >> On 15.03.21, Igor Munkin wrote: >>> Sergey, >>>=20 >>> This is ridiculous: you split two similar renames required by the = one >>> issue into *two* separate commits, but leaving the changes *totally >>> unrelated to each other* within a single commit. Please, split this >>> patch into two: one for the test directory tweak and another with >>> mocking environment variable in CMake. >>=20 >> Sorry, but why then you said nothing about commits separation/joining >> for this [1] draft series, as I asked for? >=20 > I mentioned it here[1] and you decided to left everything intact = (that's > OK for me). This commit is a new one and you created it from two > separate commits in the previous version. So it's just a new feedback > for the new changes. >=20 >>=20 >> Offline we came to the agreement that we should use 3 commits for = each >> test suite. The first to onboard it, the second to adjust for LuaJIT, >> the third to adjust it for Tarantool. >=20 > I remember only 3 bullets on which we have explicitly agreed: > * The suite should be taken intact in the first commit despite the = fact > tests can be broken. > * LuaJIT related changes should be separated from Tarantool related > ones to ease their further maintenance. Furthermore, every change > should be mentioned in our GitHub queue. > * All tests should be run in a unified way via both LuaJIT and = Tarantool > testing machinery. It means test chunk or test case is either run or > not despite the testing environment. >=20 > Regarding everything else I was open to discuss it on review, since = it's > too hard (at least for me) to consider all issues faced while adopting > these suites and create the most unified way to handle them. I = believe, > this is the purpose of review process: to see what is done and think > whether it can be done in a better way (better doesn't mean *you* made > it wrong, but *we* chose a bad solution). >=20 >> Later your asked me to send Mergens changes as is. >>=20 >> I specially send draft to discuss the commits content, order and so = on. >> You said only about separating patchset for different test suites and >> merged LuaJIT test suite, so I thought that commit order is OK for = you. >> I'd merged changes with Tarantool-related one. >=20 > I hope you remember that I've asked a couple of times to polish and = send > LuaJIT-test-cleanup patchset. The last time I was asking was right > before you sent WIP series. LuaJIT related patches per se were ready = to > be merged; the only issue was disabling strict for it. Additionally = I've > adjusted tarantool-tests runner to use LUAJIT_TEST_COMMAND to make all > tests be run by unified way. >=20 > Then I asked you to split two remaining suites into two series so I = can > look on them separately. Hence, I was OK with the general approach, = but > was afraid to miss some details. >=20 >>=20 >> If it is not comfortable for you getting WIP series and discussing = them >> let's decline this practise. >=20 > I'm totally fine with WIP series: e.g. LUAJIT_TEST_INIT has been done > after I've glanced all the patches with tests you've sent. >=20 >>=20 >> Back to business, I've split the patch into two, considering your >> proposal, see them below (their order is the same). >=20 > Nice, thanks a lot! >=20 >>=20 >=20 > >=20 >>=20 >> Patch for test directory tweak. Adjusted considering [2]. >>=20 >> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D >> commit a84980334dce27243a25508e3bb8fd491689e552 >> Author: Sergey Kaplun > >> Date: Mon Mar 15 16:24:07 2021 +0300 >>=20 >> test: adjust lua-Harness tests that using dofile >>=20 >> This patch makes out-of-source execution lua-Harness suite tests = that >> using `dofile()` correct. >=20 > Minor: There are other tests using , but they are fine. The = root > problem relates to the approach used in the patched tests: the .t = files > are kinda wrapper, containing the basic checks (or not, e.g. UTF-8 = test > chunk), and some specific checks are moved to the auxiliary chunks > "dofiled" in .t script. >=20 >>=20 >> There are the following files that used `dofile()` function >> on file in test sources directory: >> * 101-boolean.t >> * 102-function.t >> * 103-nil.t >> * 104-number.t >> * 105-string.t >> * 106-table.t >> * 107-thread.t >> * 108-userdata.t >> * 203-lexico.t >> * 231-metatable.t >> * 301-basic.t >> * 305-utf8.t >> * 404-ext.t >>=20 >> `dofile()` looks for files to execute in the current working = directory, >> that might be not the same as the test source directory. >>=20 >> This patch introduces the new function `dofile_fullpath()` that >=20 > What does "fullpath" mean? If this is absolute path, then naming is > ambiguous; and I guess nothing stops one from using relative paths = (but > not symlink, I think) here. Furthermore, the naming doesn't represent > the purpose: as I mentioned above this dofile is needed to implement > specific checks, so you can adjust the naming to more verbose one: > or . Thoughts? >=20 >> evaluates full path to file considering arg[0] (i.e. test = filename) >=20 > It's worth to check and mention the caveat with symlinks. If the issue > exists, then the paths should be resolved in CMake the way similar to > the one used for luacheck target. >=20 >> value. >>=20 >> Part of tarantool/tarantool#5844 >> Part of tarantool/tarantool#4473 >>=20 >=20 > >=20 >> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D >>=20 >> Patch for mocking environment variable in CMake. >>=20 >> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D >> commit 483508b0a7863efabcde6d232ab9af2033e0011f >> Author: Sergey Kaplun > >> Date: Mon Mar 15 22:08:07 2021 +0300 >>=20 >> test: forcify set USERNAME env var for lua-Harness >=20 > Strictly saying, you do not force, but just set. Furthermore, I doubt > about word formation you applied to "force". >=20 >>=20 >> 309-os.t checks `os.getenv()` function by examining of >=20 > Typo: examine , not examine of [2]. >=20 >> USERNAME or LOGNAME environment variable. >> These variables might not be set in the environment, that leads to = test >> failure. >>=20 >> This patchs sets manually USERNAME environment variable for >=20 > Typo: s/manually/explicitly/. >=20 >> lua-Harness-tests target. >>=20 >> Part of tarantool/tarantool#5844 >> Part of tarantool/tarantool#4473 >>=20 >> diff --git a/test/lua-Harness-tests/CMakeLists.txt = b/test/lua-Harness-tests/CMakeLists.txt >> index f8611ce..b844788 100644 >> --- a/test/lua-Harness-tests/CMakeLists.txt >> +++ b/test/lua-Harness-tests/CMakeLists.txt >> @@ -34,6 +34,11 @@ add_custom_command(TARGET lua-Harness-tests >> env >> LUA_PATH=3D"${LUA_PATH}\;" >> LUA_CPATH=3D"${LUA_CPATH}\;" >> + # XXX: 309-os.t checks os.getenv() function by examining of >=20 > Typo: examine , not examine of [2]. >=20 >> + # USERNAME or LOGNAME environment variable. >> + # These variables might not be set in the environment, so >> + # set one of them manually. >> + USERNAME=3D"fperrad" >> ${PROVE} ${CMAKE_CURRENT_SOURCE_DIR} >> --exec '${LUAJIT_TEST_COMMAND} -l profile_luajit21' >> ${LUA_TEST_FLAGS} >> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D >>=20 >> [1]: = https://lists.tarantool.org/pipermail/tarantool-patches/2021-March/022563.= html >> [2]: = https://lists.tarantool.org/pipermail/tarantool-patches/2021-March/022710.= html >>=20 >> --=20 >> Best regards, >> Sergey Kaplun >=20 > [1]: = https://lists.tarantool.org/tarantool-patches/16E39C08-397B-4A38-953A-B9EB= 85CD9C8B@tarantool.org/T/#m9de07af72e06ac22b7540741b9e1c4ac485688bb = > [2]: https://dictionary.cambridge.org/dictionary/english/examine = >=20 > --=20 > Best regards, > IM --Apple-Mail=_5EE338FA-C215-4152-AD49-8B967C912BAD Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=utf-8 Hi!

Thanks = for patch and review!

With all comments below fixed the patch is LGTM.

My proposal on = dofile_fullpath() naming:
- split the functions, = it=E2=80=99ll ease reading. Keep dofile() in the place of = use, 
  wrap the name with additional = function
- pick a name that echo what function = does: delivers a filename relative to the
  = test itself e.g. relative_to_test()
This results = in

  =   dofile(relative_to_test('lexico53/boolean.t=E2=80=99))

regards,
Sergos


On 16 Mar 2021, at 13:51, Igor Munkin <imun@tarantool.org> = wrote:

Sergey,
 
On 16.03.21, Sergey Kaplun wrote:
Igor,

Thanks for the review!

On 15.03.21, Igor Munkin wrote:
Sergey,

This is ridiculous: you split two similar renames required by = the one
issue into *two* separate commits, but leaving the = changes *totally
unrelated to each other* within a single = commit. Please, split this
patch into two: one for the = test directory tweak and another with
mocking environment = variable in CMake.

Sorry, but = why then you said nothing about commits separation/joining
for this [1] draft series, as I asked for?

I mentioned it here[1] and you decided to left everything = intact (that's
OK for me). This commit is a new one and you created it from = two
separate = commits in the previous version. So it's just a new feedback
for the new = changes.


Offline we came to the agreement that we should use 3 commits = for each
test suite. The first to onboard it, the second = to adjust for LuaJIT,
the third to adjust it for = Tarantool.

I remember only 3 bullets on which we have explicitly = agreed:
* The suite = should be taken intact in the first commit despite the fact
 tests = can be broken.
* LuaJIT related changes should be separated from Tarantool = related
 ones to = ease their further maintenance. Furthermore, every change
 should = be mentioned in our GitHub queue.
* All tests should be run in a unified way via both LuaJIT = and Tarantool
 testing = machinery. It means test chunk or test case is either run or
 not = despite the testing environment.

Regarding everything else I was open to discuss it on review, = since it's
too hard (at = least for me) to consider all issues faced while adopting
these suites = and create the most unified way to handle them. I believe,
this is the = purpose of review process: to see what is done and think
whether it = can be done in a better way (better doesn't mean *you* made
it wrong, but = *we* chose a bad solution).

Later your asked me to send Mergens = changes as is.

I specially send draft to = discuss the commits content, order and so on.
You said = only about separating patchset for different test suites and
merged LuaJIT test suite, so I thought that commit order is = OK for you.
I'd merged changes with Tarantool-related = one.

I hope you remember that I've asked a couple of times to = polish and send
LuaJIT-test-cleanup patchset. The last time I was asking was = right
before you = sent WIP series. LuaJIT related patches per se were ready to
be merged; = the only issue was disabling strict for it. Additionally I've
adjusted = tarantool-tests runner to use LUAJIT_TEST_COMMAND to make all
tests be run = by unified way.

Then I asked you to split two remaining suites into two = series so I can
look on them separately. Hence, I was OK with the general = approach, but
was afraid to = miss some details.


If it is not = comfortable for you getting WIP series and discussing them
let's decline this practise.

I'm totally = fine with WIP series: e.g. LUAJIT_TEST_INIT has been done
after I've = glanced all the patches with tests you've sent.


Back to business, I've split the patch into two, considering = your
proposal, see them below (their order is the = same).

Nice, thanks a lot!



<snipped>


Patch for test = directory tweak. Adjusted considering [2].

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
commit a84980334dce27243a25508e3bb8fd491689e552
Author: Sergey Kaplun <skaplun@tarantool.org>
Date: =   Mon Mar 15 16:24:07 2021 +0300

   test: adjust lua-Harness tests that using = dofile

   This patch makes = out-of-source execution lua-Harness suite tests that
   using `dofile()` correct.

Minor: There are other tests using <dofile>, but they = are fine. The root
problem relates to the approach used in the patched tests: = the .t files
are kinda = wrapper, containing the basic checks (or not, e.g. UTF-8 test
chunk), and = some specific checks are moved to the auxiliary chunks
"dofiled" in = .t script.


   There are the following files that used = `dofile()` function
   on file in test = sources directory:
   * 101-boolean.t
   * 102-function.t
   * 103-nil.t
   *= 104-number.t
   * 105-string.t
   * 106-table.t
   * 107-thread.t
   * 108-userdata.t
   * 203-lexico.t
   * 231-metatable.t
   * 301-basic.t
   * 305-utf8.t
   * 404-ext.t

   `dofile()` looks for files to execute in = the current working directory,
   that = might be not the same as the test source directory.

   This patch introduces the new function = `dofile_fullpath()` that

What does = "fullpath" mean? If this is absolute path, then naming is
ambiguous; = and I guess nothing stops one from using relative paths (but
not symlink, = I think) here. Furthermore, the naming doesn't represent
the purpose: = as I mentioned above this dofile is needed to implement
specific = checks, so you can adjust the naming to more verbose one:
<test_more> or <make_specific_checks>. = Thoughts?

   evaluates full path to file considering = arg[0] (i.e. test filename)

It's worth to = check and mention the caveat with symlinks. If the issue
exists, then = the paths should be resolved in CMake the way similar to
the one used = for luacheck target.

   value.

   Part of = tarantool/tarantool#5844
   Part of = tarantool/tarantool#4473


<snipped>

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D

Patch for mocking environment variable in = CMake.

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
commit 483508b0a7863efabcde6d232ab9af2033e0011f
Author: Sergey Kaplun <skaplun@tarantool.org>
Date: =   Mon Mar 15 22:08:07 2021 +0300

   test: forcify set USERNAME env var for = lua-Harness

Strictly saying, you do not force, but just set. Furthermore, = I doubt
about word = formation you applied to "force".


   309-os.t checks `os.getenv()` function by = examining of

Typo: examine <smth>, not examine of = <smth>[2].

   USERNAME or LOGNAME = environment variable.
   These variables = might not be set in the environment, that leads to test
   failure.

   This patchs sets manually USERNAME = environment variable for

Typo: = s/manually/explicitly/.

   lua-Harness-tests = target.

   Part of = tarantool/tarantool#5844
   Part of = tarantool/tarantool#4473

diff --git = a/test/lua-Harness-tests/CMakeLists.txt = b/test/lua-Harness-tests/CMakeLists.txt
index = f8611ce..b844788 100644
--- = a/test/lua-Harness-tests/CMakeLists.txt
+++ = b/test/lua-Harness-tests/CMakeLists.txt
@@ -34,6 +34,11 @@ = add_custom_command(TARGET lua-Harness-tests
  env
    LUA_PATH=3D"${LUA_PATH}\;"
    LUA_CPATH=3D"${LUA_CPATH}\;"
+    # XXX: 309-os.t checks os.getenv() = function by examining of

Typo: examine = <smth>, not examine of <smth>[2].

+ =    # USERNAME or LOGNAME environment variable.
+    # These variables might not be set in the = environment, so
+    # set one of them = manually.
+    USERNAME=3D"fperrad"
    ${PROVE} = ${CMAKE_CURRENT_SOURCE_DIR}
      --exec = '${LUAJIT_TEST_COMMAND} -l profile_luajit21'
      ${LUA_TEST_FLAGS}
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D

[1]: https://lists.tarantool.org/pipermail/tarantool-patches/2021-Ma= rch/022563.html
[2]: https://lists.tarantool.org/pipermail/tarantool-patches/2021-Ma= rch/022710.html

-- 
Best = regards,
Sergey Kaplun

[1]: https://lists.tarantool.org/tarantool-patches/16E39C08-397B-4A3= 8-953A-B9EB85CD9C8B@tarantool.org/T/#m9de07af72e06ac22b7540741b9e1c4ac4856= 88bb
[2]: https://dictionary.cambridge.org/dictionary/english/examine=

-- 
Best = regards,
IM

= --Apple-Mail=_5EE338FA-C215-4152-AD49-8B967C912BAD--