| J. Leichter. Shared tuple memories, shared memories, buses and LAN's -- Linda implementations across the spectrum of connectivity. PhD thesis, Yale University, 1989. YALEU/DCS/TR-714. |
....tuple using a dispatch and then using an arrived primitive to check if it is available. The inp primitive is not widely supported in Linda implementations. There are a number of perceived semantic problems associated with these primitives which are used as the primary reason for their removal[11], replacement[3] or, when implemented, behaviour which can potentially lead to unintentional program behaviour[1] The problem is based on the semantics of the out primitive. Given the example code fragments in Figure 7, can the inp in process two fail Some implementations[1] of Linda state that ....
....the tuple space which ts1 refers to. 13 before the [ DONE string ] is inserted, so the tuple [10 integer ] will be present. If this is not true then this allows implementations to create inp as a function that always returns false. A detailed description of the problem can be found in Leichter[11] and Rowstron[12] Here, it is assumed that out primitives are ordered. The C Linda approach of using inp has a disadvantage over the C Bonita approach. The disadvantage is the level of communication between the user process and the run time system. The C Bonita approach requires two messages to ....
J. Leichter. Shared tuple memories, shared memories, buses and LAN's -- Linda implementations across the spectrum of connectivity. PhD thesis, Yale University, 1989. YALEU/DCS/TR-714.
....matched tuple in the tuple space. rd is used for its binding and synchronization side effects. Two related operators are inp and rdp. These perform tasks equivalent to in and rd but are nonblocking. Instead they return a boolean value which indicates the success of the operation. Recent research#[10] argues against the use of these operators. 2.3 Process creation The final operation provided by Linda is the eval operation. The eval operation is syntactically similar to out except that a new process is created to evaluate each of the fields in the tuple. When the evaluation of all fields has ....
....in the tuple space is discussed in section 5. Facts correspond almost directly to standard Linda tuples. The necessity of a predicate symbol in a fact is analogous to requiring that the first field of a tuple be an actual field with a string literal value, as enforced by some Linda implementations#[10] (This requirement does not reduce the generality of the system) Formals in tuples are implemented by unbound variables. As data in Prolog is untyped (everything is a term) the data in Prolog Linda s tuples is untyped. Tuples are added to and removed from the tuple space using Prolog s database ....
[Article contains additional citation context not shown here]
J.S. Leichter, Shared Tuple Memories, Shared Memories, Buses and LAN's - Linda Implementations Across the Spectrum of Connectivity, Ph.D. Thesis, (Yale University, Yale, CT, 1989).
....tuples. The tuple is active because separate processes are spawned to evaluate each of its fields. The tuple subsequently evolves into a passive tuple resident in the tuple space. Although not proposed in the original Linda model, many implementations support two further operators, inp and rdp [22]. These are non blocking versions of in and rd which evaluate to boolean values indicating their success and, if the operation succeeds, assigns actuals to formals as before. 4.2. Platform overview Our new platform, L 2 imbo, is based on the Linda model but includes a number of significant ....
....Once again, this timing information can be used by the system to re order messages. Note that by supporting time outs on tuple space operations we are able to avoid having to provide special support for inp and rdp, the non blocking forms of in and rd found in other tuple space implementations [22]. Non blocking operations cause particular problems for implementers of distributed tuple spaces. Essentially, one must be able to satisfy the assertion that when the non blocking operation is actioned, no matching tuple is available anywhere in the tuple space, without having to lock the entire ....
J. S. Leichter, Shared Tuple Memories, Shared Memories, Buses and LAN's - Linda Implementations across the Spectrum of Connectivity, Ph.D. Thesis, Department of Computer Science, Yale University, New Haven, Connecticut, U.S. (July 1989).
....would be possible to use persistence of tuple spaces to a programmer advantage. A persistent tuple space can effectively be used as a long term store for information that will be required in the future. The first implementations of Linda were produced by Yale University and were all closed systems[Car87, Lei89, Bjo92] using precompilers to gain information about the use of tuples which could be used to distribute tuples intelligently and retrieve them quickly. Since then there have been numerous other implementations largely based on the work of [Bjo92] Distributed servers provide a mechanism for the kernel ....
J. Leichter. Shared tuple memories, shared memories, buses and LAN's -- Linda implementations across the spectrum of connectivity. PhD thesis, Yale University, 1989. YALEU/DCS/TR-714.
No context found.
J. S. Leichter, Shared Tuple Memories, Shared Memories, Buses and LAN's - Linda Implementations across the Spectrum of Connectivity, Ph.D. Thesis, Department of Computer Science, Yale University, New Haven, Connecticut, U.S. (July 1989).
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC