Remove libmemecached 1.0 requirement.

[christianchristensen]

Ref: #33

[giac]

Hi,

This pull request seems to fix the compilation of r2.1.0 on Precise but it fails on Master with a different error:

/home/god/src/php-memcached/php_memcached_session.c: In function 'ps_open_memcached':
/home/god/src/php-memcached/php_memcached_session.c:189:24: warning: assignment makes pointer from integer without a cast [enabled by default]
/home/god/src/php-memcached/php_memcached_session.c:238:54: error: 'MEMCACHED_BEHAVIOR_REMOVE_FAILED_SERVERS' undeclared (first use in this function)
/home/god/src/php-memcached/php_memcached_session.c:238:54: note: each undeclared identifier is reported only once for each function it appears in
make: *** [php_memcached_session.lo] Error 1

[auroraeosrose]

Pretty simple fix for the second issue - php_memcached_session.c is missing an ifdef on line 237

#if defined(LIBMEMCACHED_VERSION_HEX) && LIBMEMCACHED_VERSION_HEX >= 0x00049000

around that whole block fixes the compile issue

[mkoppanen]

There are quite a few issues even then, mainly missing functions (compile -Wimplicit-function-declaration).

[auroraeosrose]

It compiles with that one liner change and all the tests are passing - so either the stuff that's missing isn't tested, or something else is going on

[mkoppanen]

hmmm, i wonder how that works. Looking at the following entry from config.m4:

AC_MSG_CHECKING([for libmemcached location])
if test "$PHP_LIBMEMCACHED_DIR" != "no" && test "$PHP_LIBMEMCACHED_DIR" != "yes"; then
  if ! test -r "$PHP_LIBMEMCACHED_DIR/include/libmemcached-1.0/memcached.h"; then
    AC_MSG_ERROR([Can't find libmemcached 1.0.x headers under "$PHP_LIBMEMCACHED_DIR"])
  fi

This should fail already in configure phase if the header is not there. Is it picking up header from different installation?

[mkoppanen]

It compiles with some changes and ifdefs but still quite a few undefined symbols. It is quite a while since 1.0.x series have been release: 1.0.2 release from the 1.0 series released 2011-10-27.

I can review a patch if someone produces one. The current PR is dangerous to apply as is.

[mkoppanen]

Added preliminary support here: 47f6ff1

Passes tests with 0.44 and 0.53 for now. Not sure how stable it is, was a bit involved

[mkoppanen]

This has been fixed