On Oct 8, 2010, at 00:56, robin wrote: > From: "john gilmore" <john_w_gilm...@msn.com> > Sent: Friday, 8 October 2010 1:51 PM > >> My first post was remiss in not noting another important use of >> blanks-substring detection, which is due originally to >> the late John Cocke. > >> The most important determinant of the performance of a translator--compiler, >> interpreter, assembler, whatever--is the >> speed and efficiency with which it eliminates insignificant blanks from >> source-program text. > I'm astonished. But assuming that's true, and assuming further that translator performance should be a consideration in the design of programming languages, this impels the design of languages in which trailing blanks are never significant.
But I remember examining a code-coverage trace of a pre-TSO/E CLIST interpreter which appeared to do BALR; STM; GETMAIN; ...; FREEMAIN; LM; BR 14 for each character in the source. (I don't know whether it had previously eliminated trailing blanks.) I was motivated by the observed performance of a simple loop. -- gil
