kgu <kguento@gmail.com> wrote:
hat-trans spends a lot of time (really a lot of time) in generate
module from a module generated by alex. The original module has 330
lines of code and module generated by hat-trans has 40700 !!.
It is well-known that giving a machine-generated input to any
compiler-like tool is the best way to show up its weaknesses! I suppose
the output from Alex is very different from the kind of code that a
human would write. Since Alex creates essentially a table-driven
lexical analyser, it is possible that a large static description of the
tables is where hat-trans is spending lots of time. Also, if the
Alex-generated code is deeply nested, hat-trans may spend too much time
trying to pretty-print its output within 80 columns, when really the
formatting will not fit nicely into such a small width.
Just some ideas. If you want to send a copy of the original Alex
specification, we may be able to investigate further.
Regards,
Malcolm
_______________________________________________
Hat mailing list
Hat@haskell.org
http://www.haskell.org/mailman/listinfo/hat
I attach original specification.It's a very simple lexer specification.
I had some problem to compile module generated by hat-trans , it
spends a lot of time too and it did not compile.