Xref: utzoo comp.compilers:1470 comp.lang.fortran:4039
Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!uwm.edu!rpi!bu.edu!att!att!ima!esegue!compilers-sender
From: roman@gaudi.ccsf.caltech.edu (Roman Salvador)
Newsgroups: comp.compilers,comp.lang.fortran
Subject: time to write a compiler
Keywords: Fortran, question
Message-ID: <1990Oct31.180632.3201@nntp-server.caltech.edu>
Date: 31 Oct 90 18:06:32 GMT
Sender: compilers-sender@iecc.cambridge.ma.us
Reply-To: roman@gaudi.ccsf.caltech.edu (Roman Salvador)
Organization: California Institute of Technology, Pasadena
Lines: 16
Approved: compilers@esegue.segue.boston.ma.us


   I would like to find out the time it takes (more or less) to write a
compiler (i.e. a Fortran 90 one). Does anybody have any statistics or
experiences?. And also, about the time for the different parts of it (i.e.
lexical analysis, parsing, error checking, code generation, debugging +
testing).
   Thanks,
                Roman  (internet: roman@gaudi.caltech.edu)
[It took me about a year part-time to write an F77 compiler based on the old
Ritchie PDP-11 C compiler, but with no significant optimization beyond that
needed to compile i=i+1 into one instruction.  I suspect F90 would be
considerably harder.  By far the most work was the I/O system and all the
nit picking to catch all the special cases I forgot. -John]
-- 
Send compilers articles to compilers@esegue.segue.boston.ma.us
{ima | spdcc | world}!esegue.  Meta-mail to compilers-request@esegue.