Reference for "Lazy" Compilation?

I vaguely remember an article some years ago about an approach to
compiling where incomplete or ambiguous semantics were deferred until
later information showed up. As I recall, it was oriented toward
languages, like PL/I, where declarations of an object could potentially be
after uses of an object.


My memory of it is that as it tried to parse something, if it came to an
impasse, it deferred work on the fragment and tried another fragment. At
some point, another fragment would provide the information that the first
fragment lacked, at which point the first fragment could proceed.


The odds are high that the reference is to an ACM publication; in fact,
it's very likely to be SIGPLAN Notices. Unfortunately, I have no idea of
the title or author, so I haven't been able to locate it through the usual
indexes. I'd guess that it was published at least ten years ago, and
probably more.


Tks,
--
-- Greg Noel, Unix Guru greg@qualcomm.com or greg@noel.cts.com
--

Return to the comp.compilers page.
Search the comp.compilers archives again.