[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: warning: possibly unused local top-level variable `%foo?-procedure'
From: |
Linus Björnstam |
Subject: |
Re: warning: possibly unused local top-level variable `%foo?-procedure' |
Date: |
Fri, 07 Jun 2024 09:26:08 +0200 |
User-agent: |
Cyrus-JMAP/3.11.0-alpha0-497-g97f96844c-fm-20240526.001-g97f96844 |
Thank you for making me realize I use guile-master...
I am not really sure what that binding expands into, but it is reasonable to
think it has no effect on performance
--
Linus Björnstam
On Fri, 7 Jun 2024, at 00:52, Tomas Volf wrote:
> On 2024-06-06 22:02:50 +0200, Linus Björnstam wrote:
>> Which version are you using? All my similar issues with srfi-9 went away
>> last year when srfi things were marked maybe-unused.
>
> I am using the latest 3.0.9. However thanks to your hint about maybe-unused,
> I
> can see this:
>
> $ git log v3.0.9.. -- module/srfi/srfi-9.scm
> commit 7fef214f6e0df4004020fec530e808c476f2d2bf
> Author: Andy Wingo <wingo@pobox.com>
> Date: Sun Mar 17 09:52:49 2024 +0100
>
> Remove vestigial code from srfi-9
>
> * module/srfi/srfi-9.scm (%define-record-type): No need to define
> record-layout.
>
> commit 19c7969fff223f28cad90e21ae04a0a5852901fc
> Author: Andy Wingo <wingo@pobox.com>
> Date: Thu Aug 24 11:41:15 2023 +0200
>
> define-inlinable marks residualized procedure as maybe-unused
>
> * module/ice-9/boot-9.scm (define-inlinable):
> * module/srfi/srfi-9.scm (define-tagged-inlinable): Add maybe-unused
> declaration. Also require at least one body expr, otherwise the
> metadata declaration could escape as the proc body.
>
> So it looks like this is not in any released version yet. And indeed, when I
> try to compile it using the current master, the warning goes away.
>
> I wonder how to work around it in the mean time. I guess I could just put
> %foo?-procedure as a top-level expression. This seems to not produce any
> warnings:
>
> (define-module (x)
> #:use-module (srfi srfi-9)
> #:export (<foo>
> foo?
> make-foo))
>
> (define-record-type <foo> (make-foo) foo?)
> %foo?-procedure
>
> I guess there should be pretty much zero performance impact, correct? Are
> there
> any side effects I might not be aware of?
>
>>
>> I don't have a computer this week so I cannot try your code. Sorry.
>
> No worries, you were helpful already, enjoy your no-computer time :)
>
>>
>> --
>> Linus Björnstam
>>
>> On Wed, 5 Jun 2024, at 17:02, Tomas Volf wrote:
>> > Hello,
>> >
>> > I am getting following warning from a guild compile:
>> >
>> > warning: possibly unused local top-level variable `%foo?-procedure'
>> >
>> > And I am not sure how to tackle it. This is my full source code:
>> >
>> > (define-module (x)
>> > #:use-module (srfi srfi-9)
>> > #:export (<foo>
>> > foo?
>> > make-foo))
>> >
>> > (define-record-type <foo> (make-foo) foo?)
>> >
>> > When I try to compile it:
>> >
>> > $ guild compile -W 3 -o x.go x.scm
>> > x.scm:7:0: warning: possibly unused local top-level variable
>> > `%foo?-procedure'
>> > wrote `x.go'
>> >
>> > I would (for obvious reasons) like to keep my compilation warning-free. I
>> > can
>> > think of two approaches:
>> >
>> > 1. Export the %foo?-procedure
>> > I think this would confuse downstream users, since they are not
>> > expected to
>> > use it directly.
>> >
>> > 2. Mark the procedure as used
>> > In C I could use `(void)proc;', is there an equivalent of that
>> > construct
>> > in Guile?
>> >
>> > Are there other options? How are you approaching it?
>> >
>> > Thank you and have a nice day,
>> > Tomas Volf
>> >
>> > PS: I am not even sure why this warning happens, the `foo?' syntax
>> > transformer
>> > is exported and *does* reference it (as far as I can tell from ,expand).
>> >
>> > --
>> > There are only two hard things in Computer Science:
>> > cache invalidation, naming things and off-by-one errors.
>> >
>> > Attachments:
>> > * signature.asc
>
> Have a nice day,
> Tomas Volf
>
> --
> There are only two hard things in Computer Science:
> cache invalidation, naming things and off-by-one errors.
>
> Attachments:
> * signature.asc