Re: [Scheme-reports] read-error? and file-error? should be part of their respective packages.

1. My comment has nothing to do with any of the other procedures in the exception handling library. My problem is not with these predicates, but with the library that they are located. Are there any other examples of api in the base library for features that have been manifestly declared optional according to Section 1.3.1?
2. These 2 procedures did not exist in the sixth draft. They are new in the seventh draft. There has been no opportunity for the public to comment on the new api and syntax created between these 2 drafts. As you say yourself, "non-trivial changes now will be unfair to people currently reviewing the draft under the assumption that it is almost final", well how unfair do you believe it is that the public has had no opportunity to have input on these "non-trivial changes". The WG should provide the public at least 1 opportunity to discuss these novel features before they become finalized.

Arthur

On 11/13/12 9:32 PM, Alex Shinn wrote:

On Wed, Nov 14, 2012 at 8:17 AM, Arthur Smyles <atsmyles@x> wrote:

Well, it's not like the spec is written in stone (yet). I've thought of
the stub idea, but I believe that it is really ugly and unnecessary.
Since there is no change to the semantics, couldn't the WG have some
remedy (perhaps unanimous consent) to fix this issue before the standard
is fully petrified?

Since all other exception handling and introspection is

in base, and you want to be able to detect these even

if not using file/read yourself, I'm not sure this move

would even be desirable. We would need to vote on it.

The spec is indeed not written in stone yet, but we

need to stop at some point, and further non-trivial changes

now will be unfair to people currently reviewing the draft

under the assumption that it is almost final. It's

likely the people who think this would be a horrible

change wouldn't speak up until after the 8th draft was

published :)

--

Alex

Arthur

On 11/13/12 6:00 PM, John Cowan wrote:
> Arthur Smyles scripsit:
>
>> Both read-error? and file-error? are currently part of (scheme
>> base). Since both the read procedure and file procedures are in
>> separate libraries and are optional, it does not make sense to make
>> these 2 procedures required. I propose that read-error? be part of the
>> (scheme read) library, and that file-error? be part of the (scheme
>> file) library.
> That is an *excellent* idea, and I only wish we had thought of it.
> Unfortunately, I have to say that it just comes too late in the process.
>
> Fortunately, implementations that don't have the read and file libraries
> can easily use these stubs:
>
> (define (read-error? x) #f)
> (define (file-error? x) #f)
>

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