Bug #39571 abi_check: Does not create all prerequisites

Description:
Early on in the build process, the top-level Makefile builds the "do_abi_check" target. This target requires include/mysql_version.h, which is created by the autotools commands that precede a "make". But if you run "make clean", mysql_version.h is removed and a subsequent "make" fails.

The reason for the failure is that mysql.h includes mysql_com.h, which depends on mysql_version.h here:

#if MYSQL_VERSION_ID >= 50000
  KILL_QUERY= 254,
#endif

If mysql_version.h is not available, no KILL_QUERY line appears in the output that results when gcc processes mysql.h, and a diff failure occurs:

make abi_headers="include/mysql/plugin.h include/mysql.h" do_abi_check
make[2]: Entering directory `/src/extern/MySQL/bzr-mysql/mysql-6.0'
set -ex; \
        for file in include/mysql/plugin.h include/mysql.h; do \
                 gcc -E -nostdinc -dI \
                          -I./include \
                          -I./include/mysql \
                          -I./sql \
                          -I./include \
                          -I./include/mysql \
                          -I./sql \
                                         $file 2>/dev/null | \
                          /bin/sed -e '/^# /d' \
                                    -e '/^[     ]*$/d' \
                                    -e '/^#pragma GCC set_debug_pwd/d' \
                                    -e '/^#ident/d' > \
                                               ./abi_check.out; \
                          /usr/bin/diff -w $file.pp ./abi_check.out; \
                          /bin/rm ./abi_check.out; \
        done
+ for file in include/mysql/plugin.h include/mysql.h
+ gcc -E -nostdinc -dI -I./include -I./include/mysql -I./sql -I./include -I./include/mysql -I./sql include/mysql/plugin.h
+ /bin/sed -e '/^# /d' -e '/^[  ]*$/d' -e '/^#pragma GCC set_debug_pwd/d' -e '/^#ident/d'
+ /usr/bin/diff -w include/mysql/plugin.h.pp ./abi_check.out
+ /bin/rm ./abi_check.out
+ for file in include/mysql/plugin.h include/mysql.h
+ gcc -E -nostdinc -dI -I./include -I./include/mysql -I./sql -I./include -I./include/mysql -I./sql include/mysql.h
+ /bin/sed -e '/^# /d' -e '/^[  ]*$/d' -e '/^#pragma GCC set_debug_pwd/d' -e '/^#ident/d'
+ /usr/bin/diff -w include/mysql.h.pp ./abi_check.out
71d70
<   KILL_QUERY= 254,
make[2]: *** [do_abi_check] Error 1

How to repeat:
BUILD/compile-pentium  # this succeeds
make clean
make                   # this fails

Suggested fix:
mysql_version.h has to be created prior to do_abi_check.

The following patch fixes the problem, but it's probably best classed as a crude hack rather than the correct fix:

--- Makefile.am 2008-09-04 11:53:16 +0000
+++ Makefile.am 2008-09-21 18:58:56 +0000
@@ -291,6 +291,7 @@ abi_check_all:      $(TEST_PREPROCESSOR_HEADE
        $(MAKE) abi_headers="$^" do_abi_check
 
 do_abi_check:
+       (cd $(top_srcdir)/include && make mysql_version.h)
        set -ex; \
        for file in $(abi_headers); do \
                 @CC@ -E -nostdinc -dI \

Thank you for the report.

Verified as described.

include/mysql_version.h is generated from ./include/mysql_version.h.in 
during the configure stage. Since configure built this header it should
ideally be cleaned up during the distclean phase, as mentioned here 

http://www.gnu.org/software/automake/manual/html_node/Clean.html#Clean. 

In our case doing a make clean cleans up this header.

The following patch seemed to fix this problem.

=== modified file 'include/Makefile.am'
--- include/Makefile.am	2008-07-24 11:55:49 +0000
+++ include/Makefile.am	2008-09-27 17:55:43 +0000
@@ -15,7 +15,7 @@
 # Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
 # MA 02111-1307, USA
 
-BUILT_SOURCES =		$(HEADERS_GEN) link_sources
+BUILT_SOURCES =		link_sources
 HEADERS_GEN =		mysql_version.h my_config.h
 HEADERS_ABI =		mysql.h mysql_com.h mysql_time.h \
 			my_list.h my_alloc.h typelib.h mysql/plugin.h \
@@ -47,7 +47,7 @@
 
 
 # Some include files that may be moved and patched by configure
-DISTCLEANFILES =	sched.h $(CLEANFILES)
+DISTCLEANFILES =	$(HEADERS_GEN) sched.h $(CLEANFILES)
 
 link_sources:
 	-$(RM) -f readline openssl

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/55775

2677 V Narayanan	2008-10-08
      Bug#39571 abi_check Does not create all prerequisites
      
      The abi_check target instroduced as part of WL#4380 verifies
      changes to mysql.h. mysql.h in turn includes mysql_version.h.
      mysql_version.h is a file that is generated during the configure
      phase. We must ensure that mysql_version.h is cleaned only during
      distclean and not during clean.

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/57082

2683 V Narayanan	2008-10-27
      Bug#39571 abi_check Does not create all prerequisites
      
      The abi_check target instroduced as part of WL#4380 verifies
      changes to mysql.h. mysql.h in turn includes mysql_version.h.
      mysql_version.h is a file that is generated during the configure
      phase. We must ensure that mysql_version.h is cleaned only during
      distclean and not during clean.

Pushed into 6.0.8-alpha  (revid:v.narayanan@sun.com-20081027072328-tfofixg7zkltakzp) (version source revid:v.narayanan@sun.com-20081027083125-q2830x3v1dyqu935) (pib:5)

Pushed into 5.1.30  (revid:v.narayanan@sun.com-20081027072328-tfofixg7zkltakzp) (version source revid:v.narayanan@sun.com-20081027072328-tfofixg7zkltakzp) (pib:5)

I am still having this problem in trying to compile mysql 5.1.30 on a Kubuntu Amd64 box. Is there a more recent, patched version? If not, can anyone explain to a complete non-developer how to apply the patch mentioned in the thread?

Stefano

Hello All,

Still having the bug on a Solaris 10 compil (with gcc 3.4.3) and sources 5.1.30
When I try to apply the patch, it seems allready applied !
Any infos ?

thanks

Fix will appear in 5.1.31 and 6.0.9 per discussion with VN.

To users who cannot compile 5.1.30 (as the fix will first appear in 5.1.31) -
the most simple workaround is to run 

  make -C include

That is, if your top-level 'make' fails, run the command as above, and repeat the top-level 'make'.

Pushed into 5.1.31-ndb-6.2.17 (revid:tomas.ulin@sun.com-20090119095303-uwwvxiibtr38djii) (version source revid:tomas.ulin@sun.com-20090108105244-8opp3i85jw0uj5ib) (merge vers: 5.1.31-ndb-6.2.17) (pib:6)

Pushed into 5.1.31-ndb-6.3.21 (revid:tomas.ulin@sun.com-20090119104956-guxz190n2kh31fxl) (version source revid:tomas.ulin@sun.com-20090119104956-guxz190n2kh31fxl) (merge vers: 5.1.31-ndb-6.3.21) (pib:6)

Pushed into 5.1.31-ndb-6.4.1 (revid:tomas.ulin@sun.com-20090119144033-4aylstx5czzz88i5) (version source revid:tomas.ulin@sun.com-20090119144033-4aylstx5czzz88i5) (merge vers: 5.1.31-ndb-6.4.1) (pib:6)

Noted in 5.1.31, 6.0.9 changelogs.

The do_abi_check program run during the build process depends on
mysql_version.h but that file was not created first, resulting in
build failure.

I'm seeing this problem on GNU/Linux with 5.1.31, 5.1.36, and 5.4.1. 

Using:

./BUILD/autorun.sh
./configure
make