Re: [Sbcl-devel] rutils and sbcl 1.1.18

[Douglas K. <do...@go...>]

Sharp-backquote is defined by rutils.  It has nothing to do with backquote.

By analogy with sharp-quote being a quote of a functional form,
sharp-backquote produces functions that you don't wish to name whose
parameters you also don't wish to name other than by % and %% being two
positional parameters. Clojure has a similar concept.
James Lawrence pointed out that Clojure gets use/mention similarly wrong -
it fails to recognize when the % symbol has been quoted and will blindly
stuff in the equivalent of a gensym in a not-for-evaluation position.


On Sat, May 10, 2014 at 7:54 PM, Faré <fa...@gm...> wrote:

> On Sat, May 10, 2014 at 4:03 PM, Douglas Katzman <do...@go...> wrote:
> > Looks like we're just not getting the patch right, and that as a minor
> > hindrance, changing a function that was set-macro-character'd into a
> named
> > readtable doesn't force recompilation of files that use the readtable.
>  So
> > in part my testing has been bogus.
> >
> > To bring you up to speed: nothing has changed in SBCL in relation to
> > backquote yet, though it might. I proactively suggested to the rutils
> > maintainer that he could eliminate the assumption that SUBST was a
> > "reasonable" way to perform alpha conversion of a lambda expression by
> > instead using sb-walker in sbcl. Then I saw that my first patch - which I
> > think was correct as it stood - was inconsistent with the behavior for
> other
> > lisps.  #` is supposed to treat #`((foo) (bar)) as (progn (foo) (bar))
> which
> > I failed to realize.
> > I'll try one more time and make sure a forced recompile works.
> >
> Wait, where is that sharsign backquote thing defined???
> CLHS 2.4.8 says that it's undefined.
>
> If there's a standard for it somewhere, I'd like to know about it,
> so I may implement it in fare-quasiquote.
>
> And yes, SUBST is a very bad way of doing alpha conversion: it goes
> not only under BACKQUOTE, but under QUOTE — but not under #(), and
> maybe maybe not under `#(). In other words, it SUCKS, and programs
> that use it are probably buggy.
>
> —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics•
> http://fare.tunes.org
> You only have power over people as long as you don't take everything away
> from them. But when you've robbed a man of everything he's no longer in
> your power — he's free again. — Alexander Solzhenitsyn (1918–2008)
>