Re: [Scheme-reports] Small comments on draft 3 John Cowan (06 Aug 2011 20:50 UTC)
Re: [Scheme-reports] Small comments on draft 3 Sven Hartrumpf (08 Aug 2011 21:04 UTC)
Re: [Scheme-reports] Small comments on draft 3 John Cowan (08 Aug 2011 23:22 UTC)

Re: [Scheme-reports] Small comments on draft 3 John Cowan 06 Aug 2011 20:49 UTC

Sven Hartrumpf scripsit:

> Author block Is the non-alphabetical order of the 8 authors intended?
> Oh, you sorted on first name? That is unusual.

Yes, it's quite remarkable than when the names of the eight committee
members are read row-wise, they are in alphabetical order by first name;
but when read column-wise, they are *almost* in alphabetical order by
last name.  I have reordered them in a more standard way.

> "Pym" should be capitalized, I guess.

Fixed in trunk.

> "standardized implementation features" -> "standardized
> implementation feature identifiers"

Fixed in trunk.

> "byte" should be listed in 1.3.3.

There are others that should be here.  Editorial ticket filed.

> "By convention, the value returned by a mutation procedure is
> unspecified."  Why not returning the modified object?  This makes many
> program changes much easier.

It's too late to prescribe this now.  Most Schemes provide a unique
"unspecified object" object and return that; R6RS implementations may
return no values.

> "include-ci" seems to be very technical. Is #!fold-case not
> enough?

Include-ci makes it possible to include R5RS (or earlier) code without
touching it, and serves most of the purpose of #!fold-case (which is
defined but optional).

> "(define (make) (list ()))"
> ......................................................................
> "(set-car! s ())" No quotes?

I don't know what you are saying here.

> "fill" arguments (and on other pages, global):  Should be better named
> "obj".

I disagree; "fill" is clearer.

> "vector-fill!"  Why no optional "start" and "end"?

Substantive ticket filed.

> "partial-bytevector-copy!" -> "bytevector-copy-partial!"

The WG voted to rename all bytevector procedures to begin with
"bytevector".

> "string-map ... when the shortest list" -> map ... when the shortest
> "string-string"

Fixed in trunk.

> "vector-map ... when the shortest list" -> map ... when the shortest
> "vector-vector"

Fixed in trunk.

> "string. raise":  Please capitalize identifiers at sentence start, for
> consistency.  There might be more cases like this.

A line was dropped.  Fixed on trunk.

> "(close-port port) ...": formatting broken

Fixed on trunk.

> "the first two" -> the last two"

Fixed on trunk.

> Why are the popular abstractions "call-with-input-string" and
> "call-with-output-string" omitted?

They are trivial compositions of open-{input,output}-string and
the new call-with-port, which subsumes all such procedures.
Call-with-{input,output}-file are retained for backward compatibility.

> "before any linefeed character":  What about a carriage return
> described below?

Fixed on trunk.

> "and display is for" -> "and display for" Parallel construction.

Fixed on trunk.

> Why does "delete-file" not return the success of the operation?

R6RS compatibility.

> All lists of procedures should be sorted for ease of reference.
>
> "scheme char normalization" This looks like a "submodule" of "scheme
> char", does not it?  If so, how about "scheme char-normalization"?

The WG voted otherwise, but not conclusively, and module names are still
subject to change.

> "These features are used by cond-expand to conditionally include
> module declarations in a module." -> "These feature identifiers are
> used by cond-expand."  (Delete the "to ..." part because it is only
> half of a story that has already been told by the report.)

Fixed in trunk.

> "C memory model flags" Add a ".".

Fixed in trunk.

> "<name-version>" Can this be easily extended to version intervals?

Up to the implementation, I think.  Implementations aren't necessarily
rational numbers.

> I am tired of writing "(cond-expand ((or bigloo2.9a bigloo2.9b <and 20
> more>) ..." and extending such lists in many places several times a
> year.  (Same for other implementations, of course.)
>
> All BibTeX references should have a year attribute.  These have
> none: 1, 2, 4, 9, 17, 23

Editorial ticket filed.

--
John Cowan      <cowan@ccil.org>       http://www.ccil.org/~cowan
                Charles li reis, nostre emperesdre magnes,
                Set anz totz pleinz ad ested in Espagnes.

_______________________________________________
Scheme-reports mailing list
Scheme-reports@scheme-reports.org
http://lists.scheme-reports.org/cgi-bin/mailman/listinfo/scheme-reports