2005-01-25 21:37  seanc

	* COPYING, pmkfile: Fix up pmkfile to get around brokeness in
	  pmk(1) 0.9.0.  Part of this "fix" includes reducing the size the
	  comments (ie, license), so throw the license into COPYING.

2004-12-29 10:20  seanc

	* pgmemcache.c: Catch up with memcache(3)'s removal of a dependency
	  for a strdup(3)-like function.  Cleanup a few comments while I'm
	  here.

2004-12-24 19:00  seanc

	* README.pgmemcache, pgmemcache.c, pgmemcache.h, pgmemcache.sql.in:
	  Add memcache_server_find(/* hash */ INT4): returns the
	  hostname:port of the server assigned to the hash value.

2004-12-24 18:42  seanc

	* README.pgmemcache, pgmemcache.c, pgmemcache.h, pgmemcache.sql.in:
	  Add a new function, memcache_server_find(/* key */ TEXT).
	  Returns the hostname:port of the server that handles the given
	  key.

	  Switch to useing mcm_server_find() instead of mcm_find_server().

	  Test to make sure mcMemNewCtxt() doesn't return NULL.

2004-12-24 17:15  seanc

	* pgmemcache.h: Commit miss.

	  > Umm... the day ends in 'Y', where'd everyone go?
	  >
	  > *grin* Update the API so it compiles against libmemcache(3)
	  1.2.0.  I hate
	  > to do this, but the next version of pgmemcache will require
	  libmemache(3)
	  > 1.2.0.  Add a new memcache_flush_all() call that uses zero
	  arguments and
	  > flushes all keys for all servers in the server list.
	  >
	  > Add a memcache_flush(TEXT) call which behaves the same as
	  > memcache_flush_all(TEXT).  At some point in the future, the
	  > memcache_flush_all(TEXT) call wil grow a warning and will
	  eventually be
	  > killed off: switching to memcache_flush(TEXT) sooner rather
	  than later
	  > would be prudent.
	  >
	  > Remove nearly all references to memcache_flush_all(TEXT) in
	  favor of
	  > memcache_flush(TEXT).

2004-12-24 17:10  seanc

	* README.pgmemcache, pgmemcache.c, pgmemcache.sql.in: Umm... the
	  day ends in 'Y', where'd everyone go?

	  *grin* Update the API so it compiles against libmemcache(3)
	  1.2.0.  I hate to do this, but the next version of pgmemcache
	  will require libmemache(3) 1.2.0.  Add a new memcache_flush_all()
	  call that uses zero arguments and flushes all keys for all
	  servers in the server list.

	  Add a memcache_flush(TEXT) call which behaves the same as
	  memcache_flush_all(TEXT).  At some point in the future, the
	  memcache_flush_all(TEXT) call wil grow a warning and will
	  eventually be killed off: switching to memcache_flush(TEXT)
	  sooner rather than later would be prudent.

	  Remove nearly all references to memcache_flush_all(TEXT) in favor
	  of memcache_flush(TEXT).

2004-12-20 22:07  seanc

	* Makefile.pmk, README.pgmemcache: Update release notes to be in
	  sync with libmemcache(3) 1.1.0.  Throw in a little extra release
	  goodness (ie, a ChangeLog) while I'm here.

2004-12-20 22:06  seanc

	* pmkfile: Change debug_cflags to be just debug.  This is in sync
	  w/ what libmemcache(3) does now.

2004-12-17 16:28  seanc

	* Makefile.pmk: Add a chump release target.  Enter stage left:
	  pgmemcache 1.0rc1.

2004-12-17 16:18  seanc

	* Makefile.pmk: Add a few glob patterns to clean

2004-12-17 16:18  seanc

	* pgmemcache.h, pgmemcache.sql.in: Add $PostgreSQL$ RCS header

2004-12-17 16:05  seanc

	* README.pgmemcache: At long last, add documentation that explains
	  how pgmemcache works, is installed, is setup, the requirements,
	  the API, and also add a handful of examples.	There was much
	  rejoicing.

2004-12-17 15:20  seanc

	* LICENSE: License pgmemcache under the standard MIT license.

2004-12-17 15:18  seanc

	* Makefile, Makefile.pmk, pmkfile: Make use of pmk(1) and pgxs for
	  build infrastructure.  This should work on most in a pre-8.0
	  world, but will only work for darwin in a post 8.0rc2.

2004-12-17 14:40  seanc

	* pgmemcache.c: *) Explicitly inline the macro wrappers for
	  pfree(), palloc(), pstrdup(),    and the homegrown pstrdup().

	  *) Have the MCM_CHECK() macro now check if the global memcache
	  memory    context is NULL along with the global memcache struct.

	  *) Quench a handful of gcc(1) warnings that can't exist, but crop
	  up	because it doesn't grok that elog(ERROR,...) longjump(3)'s
	  out of the	given function.

	  *) Mechanically replace all mc_*(...) calls with mcm_*(ctxt, ...)
	  to make    use of the newly written multiple memory context API.
	  This change	 requires that libmemcache(3) 1.1.0rc2 or newer is
	  used.

	  *) Rename all MC_* macros to MCM_* to aid in detection of API
	  calls that	don't make use of the multiple memory context api.
	  There should be    nothing that matches [Mm][Cc]_* anywhere in
	  pgmemcache now.

	  *) Conditionalize the memory initialization on the global memory
	  context,    not on the global memcache struct.  In the future I'd
	  like to be able to	support multiple memcache server
	  lists/structs in the same backend.	 This aids in that eventual
	  goal.

2004-12-16 22:26  seanc

	* pgmemcache.h: libmemcache(3) header changed its name to
	  memcache.h in 1.1.0.

2004-11-30 11:19  seanc

	* Makefile, README.pgmemcache, pgmemcache.c, pgmemcache.h,
	  pgmemcache.sql.in: Initial import of pgmemcache

2004-11-30 11:19  seanc

	* Makefile, README.pgmemcache, pgmemcache.c, pgmemcache.h,
	  pgmemcache.sql.in: Initial revision

