| Bug #9166 | Cluster compile failure on Solaris 10 | ||
|---|---|---|---|
| Submitted: | 14 Mar 2005 12:39 | Modified: | 16 Mar 2005 10:52 |
| Reporter: | Kent Boortz | Email Updates: | |
| Status: | Closed | Impact on me: | |
| Category: | MySQL Server: Compiling | Severity: | S2 (Serious) |
| Version: | 4.1.10, 5.0.3 | OS: | Solaris (solaris 10) |
| Assigned to: | Tomas Ulin | CPU Architecture: | Any |
[15 Mar 2005 15:00]
Martin Skold
There is a SOLARIS define (used in mysqld.cc for example). It does not seemed to be passed to cc when compiing this file (this is a bug in the build scripts), but if this is modified, then the #define _XOPEN_SOURCE 500 can be wrappen in a #ifndef SOLARIS ... #endif However, since we probably have to modify the configure script we might as well add an -DIRIX so we directly can wrap it in #ifdef IRIX ... #endif
[15 Mar 2005 21:21]
Bugs System
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/internals/23059
[16 Mar 2005 10:52]
Martin Skold
Wrapped #define with #ifdef __sgi ... #endif
Description: Using cc-5.0 on Solaris the compile of " basestring_vsnprintf.c" fails with "/usr/include/sys/feature_tests.h", line 332: #error: "Compiler or options invalid for pre-UNIX 03 X/Open applications and pre-2001 POSIX applications" It is possible to force the code to compila if "-xc99=%none" is added to CFLAGS, but it looks like a correction for IRIX in the source file itself is causing this problem. The correction sets #define _XOPEN_SOURCE 500 before the system includes. How to repeat: Compile on Solaris 10, the source using CC=cc-5.0 CXX=CC \\ ASFLAGS="-xarch=v9" \\ CFLAGS="-DBIG_TABLES -Xa -xstrconst -mt -D_FORTEC_ -xarch=v9" \\ CXXFLAGS="-DBIG_TABLES -noex -mt -D_FORTEC_ -xarch=v9" \\ LDFLAGS="-xarch=v9" \\ ./configure --with-extra-charsets=complex --enable-thread-safe-client \\ --enable-local-infile --disable-shared --with-readline --with-ndbcluster Suggested fix: The IRIX correction in the source file should probably be controlled from configure.in or something, and be conditional for IRIX or operating systems with this problem. But I can't really say.