Sergei Golubchik added a comment - 2014-10-01 23:35 - edited So, there patches are: 1_hurd.patch we don't have hurd to test it and I don't understand why it's doing what it's doing (besides mtr fix, that's clear 33_scripts__mysql_create_system_tables__no_test.patch we might want to disable test account, but requires a discussion. I'll ask 38_scripts__mysqld_safe.sh__signals.patch this patch adds kill -1 and kill -15 support to mysqld_safe script. But there is no explanation why, besides the original comment in bug#9114 saying that "it would be so simple to enable it" 41_scripts__mysql_install_db.sh__no_test.patch followup for the 33_* patch, not to be used alone 50_mysql-test__db_test.patch ditto 62_disable_tests.patch this disables tests (disabled.def) that apparently failed somewhere for someone. I cannot do that, at least we need a bug report with the failure log. Instead debian should, perhaps, look at the failures, perhaps they're long gone 70_mysql_va_list.patch already applied, back in 2011 71_disable_rpl_tests.patch more tests in disabled.def 72_fix_standalone_tests.patch there's no explanation again. Why should mtr look in lib/mysql-testsuite/suite ? What is "standalone tests"? A separate debian package that installs additional mtr tests? 93_kfreebsd_tests.patch it's MDEV-5707 94_spelling.patch ok, will do fix-func_math-test-failure.patch cannot repeat the failure that this patch is supposed to be fixing fix-mips64el-ftbfs.patch as we don't have mips ourselves, I can only apply patches that make sense. I've applied one change in this patch (s/=h/=d/ because gcc says that "h" is no longer supported) and didn't apply the other ("=lc" makes no sense to me, the stackoverflow page talks about x86 and "I" not mips and "l") fix-mysqlhotcopy-test-failure.patch not needed, already fixed differently regex_plus.patch not needed, already fixed differently versioned_symbols.patch we already have versioned symbols

Otto Kekäläinen added a comment - 2014-10-02 08:53 Thanks, with this as backbone I will be able to remove most of the patches. While you are at it, could you quickly look at http://anonscm.debian.org/cgit/pkg-mysql/mariadb-10.0.git/tree/debian/patches and http://anonscm.debian.org/cgit/pkg-mysql/mysql-5.6.git/tree/debian/patches if they contain anything that adds value to the list above. I suspect that mysql-5.6 has lots of legacy patches and nobody from Oracle has really reviewed them for a long time, but they are easier to ignore if you confirm this stance...

Sergei Golubchik added a comment - 2014-10-02 13:04 I've moved that to a separate task, MDEV-6828