Re: [Guile-commits] GNU Guile branch, syncase-in-boot-9, updated. e3c5df539640a36eb1493f581087d54a4714f337

Re: [Guile-commits] GNU Guile branch, syncase-in-boot-9, updated. e3c5df539640a36eb1493f581087d54a4714f337

[Ludovic Courtès]

Hi!

"Andy Wingo" <wingo@pobox.com> writes:

> commit 34f3d47df9311852ba7eab6f8d1c36535c3774dd
> Author: Andy Wingo <wingo@pobox.com>
> Date:   Thu May 28 14:49:33 2009 +0200

[...]

>     * module/ice-9/boot-9.scm: Remove #' read-hash extension, which actually
>       didn't do anything at all. It's been there since 1997, but no Guile
>       code I've ever seen uses it, and it conflicts with #'x => (syntax x)
>       from modern Scheme.

I find this point questionable.  Do we really need to break this
functionality?  It would be nicer if we could both keep
`read-hash-extend' and the functionality you want.  What do you think?

(Besides, this commit appears to contain unrelated changes.  Also, I
prefer it when the test case appears in the same commit as the feature
it tests, for easier review.)

You may also want to add "SRFI-62 support" in `NEWS'.

Thanks,
Ludo'.

Re: [Guile-commits] GNU Guile branch, syncase-in-boot-9, updated. e3c5df539640a36eb1493f581087d54a4714f337

[Andy Wingo]

Howdy,

Thanks for the review!

On Sun 31 May 2009 01:52, ludo@gnu.org (Ludovic Courtès) writes:

>> commit 34f3d47df9311852ba7eab6f8d1c36535c3774dd
>> Author: Andy Wingo <wingo@pobox.com>
>> Date:   Thu May 28 14:49:33 2009 +0200
>
> [...]
>
>>     * module/ice-9/boot-9.scm: Remove #' read-hash extension, which actually
>>       didn't do anything at all. It's been there since 1997, but no Guile
>>       code I've ever seen uses it, and it conflicts with #'x => (syntax x)
>>       from modern Scheme.
>
> I find this point questionable.  Do we really need to break this
> functionality?  It would be nicer if we could both keep
> `read-hash-extend' and the functionality you want.  What do you think?

`read-hash-extend' is still there. This just removed the #'
read-hash-extension, which was the identity extension: #'foo == foo.

`read-hash-extend' overrides any core hash forms, so this extension
prevented #'foo == (syntax foo). It would be easy for an affected reader
to add this particular extension back, though.

> (Besides, this commit appears to contain unrelated changes.

They are all described by the commit log, right? It is true that some of
the changes are not related to others, though. I'll try to separate them
out more in the future.

> Also, I
> prefer it when the test case appears in the same commit as the feature
> it tests, for easier review.)

OK.

> You may also want to add "SRFI-62 support" in `NEWS'.

Yes, and to the cond-expand forms. I have a lot of things I need to add
to NEWS before the release.

Cheers,

Andy
-- 
http://wingolog.org/

Re: [Guile-commits] GNU Guile branch, syncase-in-boot-9, updated. e3c5df539640a36eb1493f581087d54a4714f337

[Neil Jerram]

Andy Wingo <wingo@pobox.com> writes:

>>>     * module/ice-9/boot-9.scm: Remove #' read-hash extension, which actually
>>>       didn't do anything at all. It's been there since 1997, but no Guile
>>>       code I've ever seen uses it, and it conflicts with #'x => (syntax x)
>>>       from modern Scheme.
>>
>> I find this point questionable.  Do we really need to break this
>> functionality?  It would be nicer if we could both keep
>> `read-hash-extend' and the functionality you want.  What do you think?
>
> `read-hash-extend' is still there. This just removed the #'
> read-hash-extension, which was the identity extension: #'foo == foo.
>
> `read-hash-extend' overrides any core hash forms, so this extension
> prevented #'foo == (syntax foo). It would be easy for an affected reader
> to add this particular extension back, though.

I don't think we need to worry much about preserving the #' read-hash
extension, however:

(1) Doesn't #'x => (syntax x) need to apply only in the lexical RHS of
a syntax-case syntax transformer?  Perhaps then we could implement
this meaning of #'x only in this limited scope, and retain the older
meaning in other scopes?

(2) We should of course mention the removal of the old #' in NEWS, and
as part of that entry it would be good to say in detail how "It would
be easy for an affected reader to add this particular extension
back".  (Which I assume just means providing the necessary
(read-hash-extend ...) incantation.)

Regards,
        Neil

Re: [Guile-commits] GNU Guile branch, syncase-in-boot-9, updated. e3c5df539640a36eb1493f581087d54a4714f337

[Andy Wingo]

Hi,

On Wed 03 Jun 2009 00:35, Neil Jerram <neil@ossau.uklinux.net> writes:

> Andy Wingo <wingo@pobox.com> writes:
>
>>>>     * module/ice-9/boot-9.scm: Remove #' read-hash extension, which actually
>>>>       didn't do anything at all. It's been there since 1997, but no Guile
>>>>       code I've ever seen uses it, and it conflicts with #'x => (syntax x)
>>>>       from modern Scheme.
>>>
>>> I find this point questionable.  Do we really need to break this
>>> functionality?  It would be nicer if we could both keep
>>> `read-hash-extend' and the functionality you want.  What do you think?
>>
>> `read-hash-extend' is still there. This just removed the #'
>> read-hash-extension, which was the identity extension: #'foo == foo.
>>
>> `read-hash-extend' overrides any core hash forms, so this extension
>> prevented #'foo == (syntax foo). It would be easy for an affected reader
>> to add this particular extension back, though.
>
> I don't think we need to worry much about preserving the #' read-hash
> extension, however:
>
> (1) Doesn't #'x => (syntax x) need to apply only in the lexical RHS of
> a syntax-case syntax transformer?  Perhaps then we could implement
> this meaning of #'x only in this limited scope, and retain the older
> meaning in other scopes?

Yes, although you can't determine that before macroexpansion, so it's
akin to `unquote' outside of `quasiquote'.

> (2) We should of course mention the removal of the old #' in NEWS, and
> as part of that entry it would be good to say in detail how "It would
> be easy for an affected reader to add this particular extension
> back".  (Which I assume just means providing the necessary
> (read-hash-extend ...) incantation.)

Yes. I'll get on the NEWS this weekend.

Cheers,

Andy
-- 
http://wingolog.org/

Re: [Guile-commits] GNU Guile branch, syncase-in-boot-9, updated. e3c5df539640a36eb1493f581087d54a4714f337

[Neil Jerram]

Andy Wingo <wingo@pobox.com> writes:

>> (1) Doesn't #'x => (syntax x) need to apply only in the lexical RHS of
>> a syntax-case syntax transformer?  Perhaps then we could implement
>> this meaning of #'x only in this limited scope, and retain the older
>> meaning in other scopes?
>
> Yes, although you can't determine that before macroexpansion, so it's
> akin to `unquote' outside of `quasiquote'.

Fair enough, I thought that might be the case.

>> (2) We should of course mention the removal of the old #' in NEWS, and
>> as part of that entry it would be good to say in detail how "It would
>> be easy for an affected reader to add this particular extension
>> back".  (Which I assume just means providing the necessary
>> (read-hash-extend ...) incantation.)
>
> Yes. I'll get on the NEWS this weekend.

Great!

Regards,
        Neil