Errata for hyperspec?

Faré fahree at gmail.com
Thu Dec 14 21:05:05 UTC 2017


>> http://www.cliki.net/CLHS%20Errata
>> Don't forget this page:
>> http://www.cliki.net/Proposed%20ANSI%20Revisions%20and%20Clarifications
>
> Actually, I created "CLHS Errata" today by copying the contents of the
> section "Minor corrections to CLHS (errata)" in "Proposed ANSI
> Revisions and Clarifications" and re-organizing them a bit.
>
> I now see that the section "Clarifications to the current standard"
> under the latter page also contains some actual erratum, for example,
>
>   Issue ERROR-EXAMPLE: Correct the invalid IF form in the
> (informational) wargames:no-win-scenario example for ERROR.
>
> I can move all errata under this section too to the "CLHS Errata" page
> and let only proposals/clarifications remain on the other one, if
> people think it makes sense.
>
Please make each page reference the other.

Some items (e.g. the NCONC issue I reported) have both errata (use the
correct fold) and requests for change.

There are also cans of worms not on that page that a specification
successor would have to face:
* the horrible unportable mess that are pathnames.
* maybe reserving some module names for defsystem, as in (require
"asdf")? Make it case-normalizing?
* standardizing, at least as options, things like threads? continuations?
* dealing with obscure corners of the MOP.
* the interaction between backquote and #n(), as in `#5(a ,@'(b c))

In a lot of cases, being able to delegate parts of the standard to
other standard/committee/document would be important: pathnames would
be defined in terms of the underlying OS so that at least all
implementations on a given platform may agree. The meaning of ASDF or
other newly "standard" modules would be delegated to their respective
authors. Threads, continuations, source-locations, debugger
interfacers or other low-level extensions would be delegated to their
respective groups, etc.

Note that standardization is not necessary for things that can be
built *on top* of CL as libraries, only for things that need to be
provided as basic interfaces not locally expressible in terms of
previous code (sure, through global transformation of a CL-in-CL
compiler, you can do everything). That's already a lot. And probably
won't be done - there are no strong stable forces driving it.

—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org
Taxonomy is the death of science — A. N. Whitehead



More information about the pro mailing list