also on the multi-version approach. but the particular reason I mentioned this was that it also tried to address the issue of composing transactions (discussion of which is explicitly excluded from the section 1.2 you refer to). for instance, compare the discussion and examples in "Composable Memory Transactions" with the one in section 3.11 (pp 80-) of Reed's thesis. not that STM isn't nicer, more complete, and all that;-) but Reed's ideas seemed rather more advanced than I had expected from the discussion of related work in the STM paper. Claus
Interesting reference. I had never heard of it. From reading section 1.2 it sounds like an early description of > the optimistic approach to implementing atomic transactions (which is itself a well-studied field).
Simon
| NAMING AND SYNCHRONIZATION IN A | DECENTRALIZED COMPUTER SYSTEM | | SourceTechnical Report: TR-205 | Year of Publication: 1978 | Author D. P. Reed | | I'm not entirely sure where I got my version from (it was mentioned | as a cornerstone in Alan Kay et al s latest project, Croquet, on which | Reed is a collaborator: http://www.opencroquet.org/ ), but here is | the abstract: | | http://portal.acm.org/citation.cfm?coll=GUIDE&dl=GUIDE&id=889815 | | (note that it mentions both grouping of updates, and synchronized | composition of modules with local synchronization constraints) | | and this might be the official site for the scanned copy (?): | | http://www.lcs.mit.edu/publications/specpub.php?id=773