FWIW, this came up recently in a discussion among a small group of people because I ran into the same problem. Sam is thinking again about start-up size -- Matthias
On Nov 4, 2014, at 1:32 PM, Manfred Lotz <manfred.l...@arcor.de> wrote: > On Tue, 04 Nov 2014 10:32:01 -0500 > Vincent St-Amour <stamo...@ccs.neu.edu> > wrote: > >> FWIW, the same program, using `racket/base` yields an executable >> that's 800k. >> >> My hypothesis is that the `typed/racket/base` version of the >> executable brings in the entire Typed Racket implementation and its >> dependencies. It probably would be sufficient to only bring in the >> run-time portions of TR, which are smaller. >> >> Would that behavior make sense for `raco exe`? >> > > Yes, of course. This makes sense. > > However, it is not really nice to have some 12MB for such an executable > where the non typed one is around 800k. > > I think this should be improved upon. > > -- > Manfred > > > >> Vincent >> >> >> >> At Tue, 4 Nov 2014 06:23:33 +0100, >> Manfred Lotz wrote: >>> >>> Hi there, >>> I have this minimal example: >>> >>> #lang typed/racket/base >>> >>> (: hello : String -> Void) >>> (define (hello f) >>> (displayln f)) >>> >>> (hello "world") >>> >>> >>> raco exe mytest.rkt >>> >>> gives an executable with size 12887089. >>> >>> >>> This doesn't seem to be ok. >>> >>> >>> -- >>> Manfred >>> >>> >>> ____________________ >>> Racket Users list: >>> http://lists.racket-lang.org/users >> ____________________ >> Racket Users list: >> http://lists.racket-lang.org/users >> > > > > ____________________ > Racket Users list: > http://lists.racket-lang.org/users ____________________ Racket Users list: http://lists.racket-lang.org/users
