databases/mariadb100-server: Update to 10.0.21

[brnrd]

Summary

The MariaDB project released version 10.0.21
Proposed commit-log:

databases/mariadb100-server: Update to 10.0.21

  - Update to 10.0.21
  - Updates mariadb100-client as well (slave-port)
  - Silence portlint
  - Re-roll patches with makepatch

[1]	https://mariadb.atlassian.net/browse/MDEV-7398
[2]	https://mariadb.atlassian.net/browse/MDEV-8128

Changes:	https://mariadb.com/kb/en/mariadb/mariadb-10021-changelog/

Differential_revision:	https://reviews.freebsd.org/D2771
Reviewed_by:	koobs (mentor), vsevolod (mentor)
Approved by:	vsevolod (mentor)
PR:		200097
Security:	36bd352d-299b-11e5-86ff-14dae9d210b8
MFH:		2015Q3

Test Plan

• testport (poudriere OK)
  mariadb100-client-10.0.20.log359 KBDownload
  mariadb100-server-10.0.20.log1 MBDownload
  -client
• portlint: makepatch and preserving the timestamps leads to many warnings

WARN: Makefile: OQGRPAH appears in PORT_OPTIONS:M, but is not listed in OPTIONS_DEFINE.
WARN: Makefile: possible use of absolute pathname "/var/db/mysql".
WARN: Makefile: for new port, make $FreeBSD$ tag in comment section empty, to make SVN happy.
WARN: Makefile: unless this is a master port, PORTNAME has to be set by "=", not by "?=".
WARN: Makefile: unless this is a master port, COMMENT has to be set by "=", not by "?=".
0 fatal errors and 5 warnings found.

• MariaDB test suite (without debug build)

Only  1978  of 3051 completed.
--------------------------------------------------------------------------
The servers were restarted 430 times
Spent 1526.127 of 2212 seconds executing testcases

Too many failed: Failed 10/1249 tests, 99.20% were successful.

Failing test(s): main.ctype_utf8 main.ctype_utf8mb4 maria.maria main.openssl_6975 main.myisam main.kill-2 main.ctype_utf16 main.ctype_utf16le main.ctype_utf32 main.ctype_utf8mb4_myisam
...
678 tests were skipped, 81 by the test itself.

mysql-test-run: *** ERROR: there were failing test cases

[brnrd]

Change email-address to brnrd@

[vsevolod]

databases/mariadb100-server/Makefile

16

You should also use your FreeBSD address (and update that for all your ports as a single commit preferably).

databases/mariadb100-server/files/patch-mysys_my__default.c

18

This change is very strange.

26

I don't understand this change. Is it legal?

databases/mariadb100-server/files/patch-storage_tokudb_ft-index_cmake__modules_TokuFeatureDetection.cmake

14

Does it work on 8.x and 9.x?

[brnrd]

Only add databases/mariadb100-server patches!

[vsevolod]

databases/mariadb100-server/Makefile

16

Could you create a separate commit that would update all ports you maintain simultaneously?

[brnrd]

Poudriere build log of mariadb100-server

mariadb100-server-10.0.19.log1 MBDownload

Poudriere build log of mariadb100-client

mariadb100-client-10.0.19.log358 KBDownload

% portlint -AC
WARN: Makefile: OQGRPAH appears in PORT_OPTIONS:M, but is not listed in OPTIONS_DEFINE.
WARN: Makefile: possible use of absolute pathname "/var/db/mysql".
WARN: Makefile: for new port, make $FreeBSD$ tag in comment section empty, to make SVN happy.
WARN: Makefile: unless this is a master port, PORTNAME has to be set by "=", not by "?=".
WARN: Makefile: unless this is a master port, COMMENT has to be set by "=", not by "?=".
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-client_CMakeLists.txt: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-cmakejemalloc.cmake: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-CMakeLists.txt: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-extra_CMakeLists.txt: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-extra_yassl_taocrypt_src_integer.cpp: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-include_CMakeLists.txt: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-libmysql_CMakeLists.txt: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-libservices_CMakeLists.txt: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-man_CMakeLists.txt: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-mysys_mydefault.c: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-pcre_CMakeLists.txt: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-scripts_CMakeLists.txt: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-scripts_mysqldsafe.sh: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-sql_CMakeLists.txt: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-sql_sqltrigger.cc: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-sql_sqlview.cc: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
WARN: /usr/freebsd-ports/databases/mariadb100-server/files/patch-sql_sysvars.cc: patch was not generated using `make makepatch''. It is recommended to use `make makepatch'' to ensure proper patch format.
0 fatal errors and 22 warnings found.

That makes NO sense... Backported all the timestamps as requested earlier, please advise!

[brnrd]

databases/mariadb100-server/Makefile

16

The only other port I'm the maintainer for is mariadb-server/client. That would now only be possible using a separate tree I'd guess. Any pointers?

[freebsd_com.jkkn.dk]

Any news / status on this ? - how can we help ? and should we look at 10.0.20 which includes the FreeBSD segmentation crash patch as well?

[koobs]

@brnrd

1. Are those makepatch portlint errors a result of your changes, or did they exist prior? Your review summary does not make that clear
2. What was the state of the mysql test suite results prior to you changes? What is the delta of failed tests between versions?

Also, you did not state the rationale for changing %%PREFIX%% to ${LOCALBASE} for the pkg-message.

The convention is to use variable substitution, as not all users know what LOCALBASE is, and even if they did, it's an implementation detail that doesnt necessarily need to be exposed to the user.

Thoughts? Comments?

[brnrd]

• Update patchset to 10.0.20
• Re-roll patches

[brnrd]

Fix LDCONFIG problem

[brnrd]

Discovered a problem with the pkg. Due to both -client and -server requiring ldconfig now there is a conflict on /usr/local/libdata/ldconfig/mariadb during install. -client installs libmysqlclient.so.18 there and -server libmysqld.so.18

[brnrd]

• Fix lib/mysql install conflict
• remove double USE_MYSQL

[brnrd]

databases/mariadb100-server/files/patch-mysys_my__default.c

17

Inherited from the mariadb55-server port

25

check the pkg-message, "MariaDB respects hier(7) and doesn't check /etc and /etc/mysql for my.cnf"
again, inherited from 5.5 port

[junovitch]

databases/mariadb100-server/Makefile

16

A minor pedantic note but one worth adding out since the spelling needs to corrected anyway... The proper capitalization would be '@FreeBSD.org' vice the lower case '@freebsd.org'. The lower case has only shows up 5 times in ports while the proper capitalization shows up 19987 times.

[koobs]

databases/mariadb100-server/Makefile

16

+1

26

1. Leaving commented entry in?
2. If this port provides a library then afaik USE_LDCONFIG is a requirement
3. What is the conflict if each of -client and -server ports install unique files in the directory? If it's the directory itself, dont @dir entries in the pkg-plist handle this already? I would get advice from portmgr on this and (2).

30

Description says 'arm', but BROKEN only checks armv6, what about other ARM versions? Update either BROKEN_* to cover them, or update description to be specific about which versions are BROKEN.

64

Could this be moved to sanity.mk given it looks like a deprecation check/warning?

If so, other MySQL/MariaDB ports should do the same. You could submit these changes in a separate revision

If not, then the message should be rewritten to explain not just that the variable is set, but why setting it is not a good thing, not required, deprecated, or whatever

138

I looked for a <OPT>_IGNORE helper in bsd.options.mk but couldnt see one. I was going to suggest using a helper here instead. We should create an <OPT>_IGNORE if its doable. You can create a new revision for this

databases/mariadb100-server/files/patch-scripts_mysql__config.sh

7

Should these have changed from substitutions to hardcoded /usr/local?

If so, why?

If not, fix

9

Should these have changed from substitutions to hardcoded /usr/local?

If so, why?

If not, fix

11

Should these have changed from substitutions to hardcoded /usr/local?

If so, why?

If not, fix

databases/mariadb100-server/files/patch-storage_tokudb_ft-index_cmake__modules_TokuFeatureDetection.cmake

14

Why doesn't the upstream source (without patching) find libexecinfo if its installed/available when this port is built?

Prefer a non-patch to fix this, and a bug report if its an upstream cmake build system issue.

databases/mariadb100-server/files/pkg-message.in

9

Should these have changed from substitutions to hardcoded /usr/local?

If so, why?

If not, fix

[brnrd]

All in all lots of legacy in here. Want to get this in for the current vulnerability http://www.vuxml.org/freebsd/36bd352d-299b-11e5-86ff-14dae9d210b8.html

Will pick up the work with Sergei from MariaDB to clean up the patches.

databases/mariadb100-server/Makefile

30

Seems I added this myself for 5.5.41 but can't find any references as to why this would indeed be broken.

64

Dates back to 2003 https://bugs.freebsd.org/56252
Can't reproduce that behaviour, remove

databases/mariadb100-server/files/patch-mysys_my__default.c

17

Traced this back to MySQL 5.1 source

databases/mariadb100-server/files/patch-storage_tokudb_ft-index_cmake__modules_TokuFeatureDetection.cmake

14

I've been cooperating with MariaDB devs and intend to rework (remove if possible) all the existing patches. This is all legacy from the existing 5.5 port (and potentially mysql etc.)

[koobs]

Not all inline comments addressed (or responded to)

[brnrd]

• fix lib/mysql install conflict
• elaborate on USE_MYSQL issue

[brnrd]

Update port to 10.0.21 and really fix LDCONFIG

[brnrd]

databases/mariadb100-server/Makefile

64

Found the reason. Problem is a circular dep in -client when USE_MYSQL is installed.

[brnrd]

Final improvements

• Solve portlint issue with files/patch-CMakeLists.txt
• Solve portlint issue with UNIQUENAME