-a memcached_st that can be cloned for use in the called function (it can not
-be used directly). It also passes a result set which does not need to be freed.
-Finally it passes a "context". This is just a pointer to a memory reference
-you supply the calling function. Currently only one value is being passed
-to each function call. In the future there will be an option to allow this
-to be an array.
-.sp
-memcached_mget_execute() and memcached_mget_execute_by_key() is
-similar to memcached_mget(), but it may trigger the supplied callbacks
-with result sets while sending out the queries. If you try to perform
-a really large multiget with memcached_mget() you may encounter a
-deadlock in the OS kernel (we fail to write data to the socket because
-the input buffer is full). memcached_mget_execute() solves this
-problem by processing some of the results before continuing sending
-out requests. Please note that this function is only available in the
-binary protocol.
-.sp
-memcached_get_by_key() and memcached_mget_by_key() behave in a similar nature
-as memcached_get() and memcached_mget(). The difference is that they take
-a master key that is used for determining which server an object was stored
-if key partitioning was used for storage.
-.sp
-All of the above functions are not testsed when the \fBMEMCACHED_BEHAVIOR_USE_UDP\fPhas been set. Executing any of these functions with this behavior on will result in
-\fBMEMCACHED_NOT_SUPPORTED\fP being returned or, for those functions which do not return
-a \fBmemcached_return_t\fP, the error function parameter will be set to
-\fBMEMCACHED_NOT_SUPPORTED\fP.
+a \fBmemcached_st\fP that can be cloned for use in the called
+function (it can not be used directly). It also passes a result set which does
+not need to be freed. Finally it passes a "context". This is just a pointer to
+a memory reference you supply the calling function. Currently only one value
+is being passed to each function call. In the future there will be an option
+to allow this to be an array.
+.sp
+\fI\%memcached_mget_execute()\fP and \fI\%memcached_mget_execute_by_key()\fP
+is similar to \fI\%memcached_mget()\fP, but it may trigger the supplied
+callbacks with result sets while sending out the queries. If you try to
+perform a really large multiget with \fI\%memcached_mget()\fP you may
+encounter a deadlock in the OS kernel (it will fail to write data to the
+socket because the input buffer is full). \fI\%memcached_mget_execute()\fP
+solves this problem by processing some of the results before continuing
+sending out requests. Please note that this function is only available in
+the binary protocol.
+.sp
+\fI\%memcached_get_by_key()\fP and \fI\%memcached_mget_by_key()\fP behave
+in a similar nature as \fI\%memcached_get()\fP and \fI\%memcached_mget()\fP.
+The difference is that they take a master key that is used for determining
+which server an object was stored if key partitioning was used for storage.
+.sp
+All of the above functions are not tested when the
+\fBMEMCACHED_BEHAVIOR_USE_UDP\fP has been set. Executing any of these
+functions with this behavior on will result in \fBMEMCACHED_NOT_SUPPORTED\fP being returned, or for those functions which do not return a \fBmemcached_return_t\fP, the error function parameter will be set to \fBMEMCACHED_NOT_SUPPORTED\fP.