rw.mutex, attach.rw.mutex {bigmemory} | R Documentation |
These functions provide additional read/write mutexes (mutual exclusions) for use
with big.matrix
objects in shared memory (class big.matrix
of
bigmemory). They are probably unnecessary for most users.
rw.mutex() attach.rw.mutex(mutexId) rlock(x) rwlock(x) unlock(x)
mutexId |
the shared memory key for the mutex. |
x |
an object of type rw.mutex . |
Mutexes are provided with every shared matrix (including file-backed matrices) and are
managed transparently by the various
functions provided for big.matrix
objects (mwhich
,
for example). However, the user may create an additional layer of mutexes
using these functions. This may be important in certain shared memory applications.
Note that describe
should be used to retrieve the rw.mutex
information.
Mutexes will be stripped out of bigmemory in versions 4.0 and beyond, leaving them available (optionally) to the user in a new package, synchronicity, in the Fall of 2009. This is largely for performance reasons with bigmemory, where the vast majority of shared memory work is read-only or involves algorithms that don't risk read/write conflicts.
A new instance of type rw.mutex
is returned.
We recommend you contact us if you have questions about user mutexes, which are not for the faint of heart.
John W. Emerson and Michael J. Kane
http://www.boost.org
big.matrix
, shared.big.matrix
,
filebacked.big.matrix
, describe
.
# None.