Comment 1 Zsolt Udvari 2024-10-10 09:45:43 UTC

The "make test" fails:
===>  Testing for wt-4.11.0
/bin/sh: /wrkdirs/usr/ports/www/wt/work/.build/test/test.wt: not found
*** Error code 127

Could you please check it?

(In reply to Zsolt Udvari from comment #1)

I believe it's because of this:

CMAKE_OFF=	BUILD_EXAMPLES BUILD_TESTS ENABLE_MYSQLSERVER \
		ENABLE_QT4 ENABLE_QT5 ENABLE_QT6
CMAKE_TESTING_ON=	BUILD_TESTS

If I remove BUILD_TESTS it's going to build many more files but indeed performs the test.

However the tests themselves fail:

===>  Testing for wt-4.11.0
Running 310 test cases...
bcrypted password: $2y$07$SCXqPUnXb1fqbVbAQFnOKOu7ZdVwvDZ4qjqMsswSfI2uIUUciXxLS
verify() took: 5.786ms
unknown location(0): fatal error: in "chart_test_WDateTimeChartMinutes": memory access violation at address: 0x18: no mapping at fault address
/home/mamadou/www/wt/work/wt-4.11.0/test/chart/WChartTest.C(72): last checkpoint: "chart_test_WDateTimeChartMinutes" test entry

*** 1 failure is detected in the test module "Master Test Suite"
__cxa_thread_call_dtors: dtr 0x827afeef0 from unloaded dso, skipping
Illegal instruction (core dumped)
*** Error code 132

Stop.
make: stopped in /home/mamadou/www/wt

So, not very sure what's the best approach here. Should I remove the tests altogether?

Just for the clarification, since I did not mention from which line. I mean removing BUILD_TESTS from CMAKE_OFF

CMAKE_OFF=	BUILD_EXAMPLES ENABLE_MYSQLSERVER \
		ENABLE_QT4 ENABLE_QT5 ENABLE_QT6
CMAKE_TESTING_ON=	BUILD_TESTS

Comment 4 Zsolt Udvari 2024-10-11 05:07:44 UTC

(In reply to Mamadou Babaei from comment #2)
I would suggest the tests and if you can correct would be nice.

I've found similar error:https://stackoverflow.com/questions/14662956/boost-test-and-async-io-memory-access-violation-at-address-no-mapping-at-fau and an old bug report: https://redmine.emweb.be/boards/2/topics/5643?r=5757#message-5757
And maybe check various linuxes and BSDs about patches.

Thank you! I reported it upstream to get their input: https://redmine.emweb.be/issues/13112

I've got a response from the upstream and they found the root cause:

https://redmine.emweb.be/issues/13112#note-4

> I have been able to reproduce the issue. It is cause by the absence of any fonts on the system at all.
> Now I think this should obviously be handled cleaner by Wt (i.e. an exception/assert).

> But that would still lead to a failing test, which can be caught, but I think that's against the spirit of the test itself.

> So I suppose it's probably better to ensure that at least one font is installed as a dependency then.

I see two approaches now, as suggested we make a dependency on a default set of fons from Ports and install them. Or, patch it to disable that specific test.

What's your idea?

Bad formatting sorry, here's the quoted reply:

> I have been able to reproduce the issue. It is cause by the absence
> of any fonts on the system at all.
> Now I think this should obviously be handled cleaner by Wt
> (i.e. an exception/assert).
>
> But that would still lead to a failing test, which can be caught,
> but I think that's against the spirit of the test itself.
>
> So I suppose it's probably better to ensure that at least one
> font is installed as a dependency then.

Comment 8 Zsolt Udvari 2024-10-17 18:39:14 UTC

(In reply to Mamadou Babaei from comment #6)
The better solution is install the tests dependencies and pass every test.
If you can't find which font is needed can disable (eg. with patch) failed tests. Or a last resort can set TESTING_UNSAFE variable (https://cgit.freebsd.org/ports/tree/Mk/Features/testing.mk#n7).

Created attachment 254375 [details]
new patch

Managed to pass the failing test by adding a test dependency on roboto-fonts-ttf. But, there are other failures now.

Will reach out to upstream once more.


/home/mamadou/www/wt/work/wt-4.11.0/test/auth/PasswordServiceTest.C(140): fatal error: in "throttle_enabled_failure_with_last_attempt_offset_test": critical check f.myPasswordService_->delayForNextAttempt(user) == attemptResults[failures] has failed
Dbo.backend.Sqlite3: begin transaction
Dbo.backend.Sqlite3: drop table "auth_token"
Dbo.backend.Sqlite3: drop table "auth_identity"
Dbo.backend.Sqlite3: drop table "auth_info"
Dbo.backend.Sqlite3: drop table "user"
Dbo.backend.Sqlite3: commit transaction
Field: '1'
Field: 'a'
Field: 'b'
Field: 'a'
Field: 'b'
Field: ''1''
Field: ''1''
Field: ''Barts'' car''
Field: ''Barts'', car''
Field: 'bike'
Field: 'cast(round(number, 2) as text) AS column_number'
Field: 'region'
Field: 'product'
Field: 'SUM(quantity) AS product_units'
Field: 'SUM(amount) AS product_sales'
Field: 'a'
Field: 'b'
Field: 'c'
Field: 'from_a'
Field: 'from_b'
Field: 'a_from'
Field: 'b_from'
Field: 'from_a'
Field: 'from_b'
Field: 'a_from'
Field: 'b_from'
Field: 'from_a'
Field: 'SUM(SELECT from_b from table)'
Dbo.backend.Sqlite3: begin transaction
Dbo.backend.Sqlite3: create table "a" (
  "id" integer primary key autoincrement,
  "version" integer not null,
  "date" text,
  "date_time" text,
  "time_point" text
)
Dbo.backend.Sqlite3: commit transaction
Dbo.backend.Sqlite3: begin transaction
Dbo.backend.Sqlite3: insert into "a" ("version", "date", "date_time", "time_point") values (?, ?, ?, ?)
Dbo.backend.Sqlite3: commit transaction
Dbo.backend.Sqlite3: begin transaction
Dbo.backend.Sqlite3: SELECT date as col0, date_time as col1, time_point as col2 FROM a limit ?
Dbo.backend.Sqlite3: commit transaction
Dbo.backend.Sqlite3: begin transaction
Dbo.backend.Sqlite3: select "id", "version", "date", "date_time", "time_point" from "a"  limit ?
Dbo.backend.Sqlite3: commit transaction
Dbo.backend.Sqlite3: begin transaction
Dbo.backend.Sqlite3: create table "a" (
  "id" integer primary key autoincrement,
  "version" integer not null,
  "date" real,
  "date_time" real,
  "time_point" real
)
Dbo.backend.Sqlite3: commit transaction
Dbo.backend.Sqlite3: begin transaction
Dbo.backend.Sqlite3: insert into "a" ("version", "date", "date_time", "time_point") values (?, ?, ?, ?)
Dbo.backend.Sqlite3: commit transaction
Dbo.backend.Sqlite3: begin transaction
Dbo.backend.Sqlite3: 
          SELECT
            strftime('%Y-%m-%d', date) as col0,
            strftime('%Y-%m-%dT%H:%M:%f', date_time) as col1,
            strftime('%Y-%m-%dT%H:%M:%f', time_point) as col2
          FROM a
         limit ?
Dbo.backend.Sqlite3: 
          SELECT
            date as col0,
            julianday('1950-03-12') as col1,
            date_time as col2,
            julianday('1948-10-02T03:05:12.009') as col3,
            time_point as col4,
            julianday('1990-02-14T01:50:05.012') as col5
          FROM a
         limit ?
Dbo.backend.Sqlite3: commit transaction
Dbo.backend.Sqlite3: begin transaction
Dbo.backend.Sqlite3: select "id", "version", "date", "date_time", "time_point" from "a"  limit ?
Dbo.backend.Sqlite3: commit transaction
Dbo.backend.Sqlite3: begin transaction
Dbo.backend.Sqlite3: create table "a" (
  "id" integer primary key autoincrement,
  "version" integer not null,
  "date" integer,
  "date_time" integer,
  "time_point" integer
)
Dbo.backend.Sqlite3: commit transaction
Dbo.backend.Sqlite3: begin transaction
Dbo.backend.Sqlite3: insert into "a" ("version", "date", "date_time", "time_point") values (?, ?, ?, ?)
Dbo.backend.Sqlite3: commit transaction
Dbo.backend.Sqlite3: begin transaction
Dbo.backend.Sqlite3: 
          SELECT
            date as col0,
            strftime('%s', '1950-03-12') as col1,
            date_time as col2,
            strftime('%s', '1948-10-02T03:05:12') as col3,
            time_point as col4,
            strftime('%s', '1990-02-14T01:50:05') as col5
          FROM a
         limit ?
Dbo.backend.Sqlite3: commit transaction
Dbo.backend.Sqlite3: begin transaction
Dbo.backend.Sqlite3: select "id", "version", "date", "date_time", "time_point" from "a"  limit ?
Dbo.backend.Sqlite3: commit transaction

*** 1 failure is detected in the test module "Master Test Suite"
__cxa_thread_call_dtors: dtr 0x828e5bef0 from unloaded dso, skipping
*** Error code 201

Stop.
make: stopped in /home/mamadou/www/wt
root@fbsd-ports:/home/mamadou/www/wt #

Comment 10 Zsolt Udvari 2024-10-21 14:22:25 UTC

(In reply to Mamadou Babaei from comment #9)
For me works fine. I did `poudriere bulk -i -j JAILNAME www/wt', install the roboto-fonts-ttf, and in this jail run 'make test' (from /usr/ports/www/wt).

How did you run `make test'?

(In reply to Zsolt Udvari from comment #10)

For maintaining ports I do have a Virtual Machine specifically for that purpose and I'm not using any jails.

I just modified it with:

TEST_RUN_DEPENDS=	roboto-fonts-ttf>=0:x11-fonts/roboto-fonts-ttf

And ran:

$ make test

But, I will investigate it later if it still fails for you.

BTW, on the other DBO test failure from https://redmine.emweb.be/issues/13112#change-65705

> this is an issue I am aware of, and something that not all platforms seem to
> handle in the same way. It results from the usage of an 
> std::this_thread::sleep_for to verify authentication throttling functionality.
> Annoyingly this works fine most of the time, but since it depends on the
> scheduler, is not guaranteed. That's not good for a test obviously.
> 
> Ticket #13130 tracks it removal, and is part of 4.11.1. Making it part of
> 4.11.0 would be difficult now, as people have already started using it, and
> changing that release will lead to issues.

I believe we need to patch it for that one.

Created attachment 254423 [details]
a new patch passing all tests

This patch disabled the Wt/DBO test which is scheduled to be part of the next release v4.11.1 by upstream addressed in this commit: https://github.com/emweb/wt/commit/c1e6550451672f8c4e2a1f24caec6473261298f2

For me, it also passes the other failing test with the addition of the Roboto font.

This is the output of make test:

*** No errors detected
__cxa_thread_call_dtors: dtr 0x8279b9ef0 from unloaded dso, skipping

I can remove the Roboto font and test again to see if it fails.

So, just pkg removed roboto and removed the test dependency on the robot and again it fails for me:

====> Compressing man pages (compress-man)
===>  Testing for wt-4.11.0
Running 310 test cases...
bcrypted password: $2y$07$LRTxTSbyPhTMQkrtKTSxTuUtgylJiWkKSUzsZed.c42.Z/W3uTnBa
verify() took: 5.839ms
unknown location(0): fatal error: in "chart_test_WDateTimeChartMinutes": memory access violation at address: 0x18: no mapping at fault address
/home/mamadou/www/wt/work/wt-4.11.0/test/chart/WChartTest.C(72): last checkpoint: "chart_test_WDateTimeChartMinutes" test entry

*** 1 failure is detected in the test module "Master Test Suite"
__cxa_thread_call_dtors: dtr 0x828e58ef0 from unloaded dso, skipping
Illegal instruction (core dumped)
*** Error code 132

Stop.
make: stopped in /home/mamadou/www/wt


Re-added:

TEST_RUN_DEPENDS=	webfonts>=0:x11-fonts/webfonts

And then:

$ make test

However:

pkg info | grep roboto

Returned nothing. Which means ports is not installing it.

But, I added the following:

RUN_DEPENDS=	roboto-fonts-ttf>=0:x11-fonts/roboto-fonts-ttf
TEST_RUN_DEPENDS=	roboto-fonts-ttf>=0:x11-fonts/roboto-fonts-ttf

And then:

$ make test

Passes with flying colors.

I'll submit another patch.

Created attachment 254426 [details]
patch that passes all tests

Passed all tests

Comment 18 Zsolt Udvari 2024-10-21 15:42:58 UTC

Sorry, I forgot to mention I rewrite TEST_RUN_DEPENDS to TEST_DEPENDS - IMHO there isn't exist TEST_RUN_DEPENDS variable. If you try it I hope your test will pass too :)

Thank you for mentioning that, but I remember I saw it on some ports (as I usually look at other ports). Example:

https://lists.freebsd.org/archives/dev-commits-ports-all/2023-March/055699.html

I'll submit another patch then. Maybe it has been removed?

Created attachment 254428 [details]
patch with only test_depends

This patch removes TEST_RUN_DEPENDS and RUN_DEPENDS and replaces it with TEST_DEPENDS.

Comment 21 Zsolt Udvari 2024-10-21 18:32:28 UTC

(In reply to Mamadou Babaei from comment #19)
I don't know. I think now it doesn't exist: if I try run 'make test' with TEST_RUN_DEPENDS the port system doesn't want install it - but with TEST_DEPENDS yes.
Here is the list of *_DEPENDS: https://cgit.freebsd.org/ports/tree/Mk/bsd.port.mk#n303
And if I `grep -r TEST /usr/ports/Mk/*' I couldn't find nothing about TEST_RUN_DEPENDS.

I'll ask it on mailing list because there are 8 ports which set TEST_RUN_DEPENDS.

Thank you! Do not bother with it.

The last patch is with TEST_DEPENDS and withoutTEST_RUN_DEPENDS and it works fine

Comment 23 Zsolt Udvari 2024-10-22 18:00:08 UTC

Committed, thanks!

Comment 24 commit-hook 2024-10-22 18:00:29 UTC

A commit in branch main references this bug:

URL: https://cgit.FreeBSD.org/ports/commit/?id=ca9312dc96c81093926046227427d0c1ccd9676f

commit ca9312dc96c81093926046227427d0c1ccd9676f
Author:     Mamadou Babaei <info@babaei.net>
AuthorDate: 2024-10-22 17:55:40 +0000
Commit:     Zsolt Udvari <uzsolt@FreeBSD.org>
CommitDate: 2024-10-22 17:57:53 +0000

    www/wt: Update to 4.11.0

    Add TEST_DEPENDS
    Pet portfmt.

    Changelog: https://webtoolkit.eu/wt/doc/reference/html/Releasenotes.html

    PR:             281959
    Approved by:    submitter is maintainer

 www/wt/Makefile                                    | 83 +++++++++++-----------
 www/wt/distinfo                                    |  6 +-
 www/wt/files/patch-src_web_FileUtils.C (gone)      | 10 ---
 .../patch-test_auth_PasswordServiceTest.C (new)    | 31 ++++++++
 www/wt/pkg-plist                                   |  7 ++
 5 files changed, 83 insertions(+), 54 deletions(-)