Ets/dets iterators & match specs in R8

Jim Larson jim@REDACTED
Mon Oct 30 20:36:08 CET 2000


In message <o1zojmk2z6.fsf@REDACTED> Bjorn writes:
>For the R8 release, we have PRELIMINARY plans to improve the
>ets/dets iterators and to add compiler/parse transform support
>to generate match specs automatically for the ets:select/2 function.
>
>Here are some of the extensions we are thinking of. We might do
>some of these, or all of them, or something entirely different
>if we (or someone else) come up with better ideas.
>...

One of the nice things about ETS is that its API doesn't introduce
any new language features - it *could* be implemented purely as
a library on top of message-passing.  I'd be suspicious of any new
ETS features which would break this property.  I'd prefer to see
either:

	- an API which uses closures whenever possible to avoid
	  creating new language features, or;

	- language features which are well-defined syntatctic sugar,
	  compiling to calls which, hopefully, could be intercepted
	  by user code.  For example, allow this feature to work with
	  any data structure which has an appropriate map function.

Jim Larson



More information about the erlang-questions mailing list