Re: [PATCH] Add atomic-box-update! function to (ice-9 atomic)

[Andrew Tropin <andrew@trop.in>]

[-- Attachment #1: Type: text/plain, Size: 1621 bytes --]

On 2023-06-22 17:42, Philip McGrath wrote:

> On Thu, Jun 22, 2023, at 5:02 AM, Andrew Tropin wrote:
>> On 2023-06-22 01:21, Philip McGrath wrote:
>>>
>>> In any case, the current documentation for
>>> atomic-box-compare-and-swap! is clear that the comparison is eq?: it
>>> just means that, when the behavior of eq? is unreliable, so is the
>>> behavior of atomic-box-compare-and-swap!.
>>
>> Good.  Than implementation of atomic-box-update! looks correct to me.
>> Any thoughts on including it in (ice-9 atomic)?
>>
>>>
>>> Tangentially, does atomic-box-compare-and-swap! handle spurious
>>> failures on architectures like ARM, or does it expose machine
>>> semantics to the programmer? Maybe that's covered by details of the
>>> C11 memory model, but I don't think "sequential consistency" alone is
>>> enough to answer the question.
>>>
>
> I think your implementation is correct. If
> atomic-box-compare-and-swap! does not handle spurious failures, I
> think your implementation is still correct, but a more efficient
> implementation could retry without an extra call to the update
> procedure.

How? If value was updated you need to recalculate new value.

>
> Someone with a better sense of Guile performance might have a view
> about whether `apply` is likely to be expensive: alternatively, the
> update procedure could be restricted to a single argument, or
> `case-lambda` could provide a specialized entry-point.

Original implementation was a single argument, but I thought that
variable number of arguments will be more convinient.

-- 
Best regards,
Andrew Tropin

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 832 bytes --]