🚀 go-pugleaf

In article <4ja3tl$pd6@solutions.solon.com>, tanmoy@qcd.Lanl.GOV (Tanmoy Bhattacharya) writes:
|>
|> If I were designing a compiler, why should I spend time and effort
|> writing complicated code to recognize a quintessentially bad
|> programming construct? In this case, it turns out that the loop is
|> easy to recognize, and many compilers will probably do so: but this is
|> a very specific case.

OTOH, compilers for languages that don't have such nice structured
constructs, like Fortran, do a better job of recognizing oddly-formed
loops, because they have to.  So, if you must program with goto's,
find a C compiler whose back end is in common with a Fortran compiler.

(Tutorial for those who didn't understand "back end":  many (most?)
compilers these days are divided into a "front end" (FE) and "back end"
(BE).  The language-specific FE parses the language and produces
an intermediate language (IL) which is read by the target-machine-specific
BE.  If the IL is defined carefully, FE's for multiple languages can be
connected to BE's for multiple target machines, and you get several
compilers for the price of one.  Most optimizations are done in the BE.
More recent compilers sometimes add a "middle end" that does
optimizations independent of both language and target machine.)