Research Contributions of FT-Linda

		Dave Bakken

		Jan 3, 1994



     1.	Careful analysis of failure problems with current Linda
	applications and with using Linda for ftol distributed systems.
	Sub-result: explored exactly what atomicity Linda must provide
	to support fault-tolerance.



     2.	Clean extensions to Linda to permit ftol dist. programming.



     3.	Efficient implementation designed for #2.



     4.	Better semantics for Linda

		a. strong inp/rdp on distributed Linda		


		b. oldest matching semantics


		c. [sequential ordering guarantee]


     5.	Multiple TS implementation (I think we're the first here)
	Gelernter had a paper on this but recalled it.



     6.	Attributes for TSs.



     7.	Tuple Transfer primitives.



Obviously, Contributions 1--3 are the major ones, though 4--7
are language contributions of interest to the Linda community.