Re: [Sbcl-devel] fast-read-char

[Raymond T. <to...@rt...>]

>>>>> "Daniel" == Daniel Barlow <da...@te...> writes:

    Daniel> Any counter-opinions?  I assume that this is a fair test on the
    Daniel> grounds that building SBCL includes a reasonably large number of calls
    Daniel> to READ, but if there's some other usage model it maybe should be
    Daniel> tried with, please pipe up ...

Wouldn't a better test be to just suck in some large file and see what
happens?

I get the feeling with CMUCL that it doesn't spend all that much time
reading files in but does spend a fair amount in actually compiling
it.  If SBCL's compiler is significantly slower than CMUCL, then the
timings will be dominated by compiling and not reading.  Especially
since without fast-read-char, compiling was *faster* than with. :-)

Ray