Page 2 of 2

Re: Macros considered harmful

Posted: Mon Apr 12, 2010 11:16 am
by Jasper
It seems to me that the real main thing here is worries about the complexity of the black boxes below you is being taken too far.. I mean, i have yet to encounter any bug in sbcl, and even if i did, i could switch implementation/report it, try to fix it. And there will always be the black box of the processor below that. Really i don't care if my functions are translated to machine code or Swahili, as long as it behaves equivalently, which it does.

Of course i do care about the ease of use and simplicity of what i end up working with, but this doesn't seem this discussion.

Of course simplicity and neatness can be nice things to have in a project, especially if you're getting mired, and if you have good insight into some project, it can sometimes be a good idea to see if you can tear bits apart and put them back again in a better ways. Both because it can simplify, and because the ability to do that indicates that it is compartmentalized. Neatness and simplicity also helps make it easier for other people to help. (But not sure if many outside people help with sbcl and such.)

Further, computers might be getting smaller, but i still don't want to sacrifice how small a computer i can run stuff on, nor how fast i can run on it. If speed was an issue atm, i will always look at the *factor* the implementation adds, not *the absolute change*.

Edit: Not really reacting about picolisp here. Of course we should be open about which implementation we use.. It would be neat if we could formalize the tests of various projects, and make it easy to compile/interpret with the various implementations, that would arguably make for a very good benchmark. (And consistency.) Unfortunately never found something like :test-components for asd, and projects seem to use their own little stuff.. Edit: okey, picolisp is another dialect.. sorry for not noticing..