Re: [boost] Latest version of Lexertl

From: Hartmut Kaiser (hartmut.kaiser_at_[hidden])
Date: 2007-11-21 13:59:16

Ben,

> >> Yes, but as we said (unless you sent this before your other
> >> reply) a 4.25MB (0x110000 * 4) lookup is definitely 
> unacceptable.  I 
> >> suppose for the case of Windows (a 256K lookup - 65536 * 
> 4) this is 
> >> OK, so we could consider an auto compression scheme if you exceed 
> >> 65536 entries for the vector in question.  Do you think 
> even 256K is 
> >> too much?
> >
> >2^16 (64k) is probably the max you should use to cut off if larger.
> Remember
> >the cache is not only there for the data, instructions are held in
> there as
> >well... 
> >Can you make this max size configurable?
> 
> Sure, I'll need to understand how to do the compression first though.

Sure.

> >One last note: parser/tokeniser/str_token.hpp still includes 
> >size_t.h/consts.h (it should include size_t.hpp/consts.hpp), 
> same for 
> >examples/cpp_code.hpp and examples/csharp_code.hpp.
> 
> My bad: just delete the directory tree (lexer) and 
> re-unzip...  those files aren't used anymore.

Ohh, that's too bad, because I'm using the cpp_code.hpp and serialise.hpp
headers in the lexertl based lexer for Wave. Any chance to revive these?

Regards Hartmut