On Mon, 22 Jun 2026, Paul Koning via cctalk wrote:
Wow. I've never heard of that, or seen it done. And I would certainly never
have considered doing so myself.
Then again, one of my DEC colleagues wrote a parser in COBOL, but that was more a
"watch this -- hold my beer" type of thing.
Wow
Was there also a wager involved?
It used to be traditional in graduate school compiler construction
classes, that one had to develop it until the compiler for a language was
written in that language and compiled in the final iteration of itself.
While being a traditional test of certain aspects of completeness, it
pointedly ignores the fact that any language has an "orientation" of its
own. Some languages are not optimized towards writing compilers, and will
require unnecessary amounts of work, and may result in a compiler that is
not as good as one written in a language oprimized for that sort of
programming.
Would a COBOL, FORTRAN or RPG compiler written in COBOL, FORTRAN or RPG be
as good as one written in a slightly lower level language? and how long
would somebody have to hold that beer? (room temperature, flat, and
stale?)
How many compilers written in the language that it compiles are as good
as one written in carefully crafted assembly language?
--
Grumpy Ol' Fred [email protected]
