On Tue, 30 Mar 2004 14:44:32 -0300, Humberto Massa said:
> There is another fail in your reasons here: as I said before, it just
> _happens_ to happen that fw[] = {} *is* the source code. What we must
> decide is what to do in the cases where *we don't know*.
>
> After all, what happens is somebody *actually* prefers the binary blob
> for modification? And, what happens if _the copyright holder_ actually
> prefers the binary blob for modification? No inconsistency here.In fact, I was going to suggest that quite possibly, there really *is* no other option for source, because the target hardware doesn't have a functional enough toolchain, so hand-assembly really *is* the most reasonable thing. That file isn't any worse than my early '80s Compiler Design class project (done the last year before lex/yacc became available for the class) - one file was a large 247x139x3 (or thereabouts) array of integers, encoding all the shift/reduce productions for a compiler for a Pascal subset that actually generated working (though painfully ugly) code for the IBM S/370 architecture. (The assignment was "Pascal subset plus your choice of one extension" - my teammate and I made the mistake of choosing "Fortran-style mixed-mode arithmetic", that array was literally 1/4 the size before we started that.) "You are lost in a maze of tiny little shift/reduce productions, all nearly alike" And yes, the two of us generated the whole table by hand. I admit we did finally break down and write some tools to verify that each line of the table did in fact have the right number of entries, and to add a column of zeros when the number of productions went up. *Preferred* source format? lex/yacc input files. Did the source ever actually *exist* in that format? Nope... :)
pgpvpecSKbqt6.pgp
Description: PGP signature
