Fix strtod subnormal rounding (bug 30220)

[Carlos O'Donell]

On 8/19/24 11:00 AM, Joseph Myers wrote:
> On Mon, 19 Aug 2024, Carlos O'Donell wrote:
> 
>> On 8/16/24 6:25 PM, Joseph Myers wrote:
>>> This patch is relative to a tree with
>>> <https://sourceware.org/pipermail/libc-alpha/2024-August/159229.html>
>>> (pending review) applied.  There is no logical dependency on that
>>> patch; it's simply that both of them involve regenerating
>>> tst-strtod-round-data.h.
>>
>> While there is no logical dependency, as a reviewer I want to try apply 
>> all of your patches and review and test together to make my review more 
>> meaningful.
>>
>> Would it be possible for you to post these together as a "loose" series 
>> that collects everything for pre-commit CI to test and a reviewer to 
>> apply?
> 
> The "series" is ongoing in that there are still more gaps in test coverage 
> to fill, with each patch being posted when ready, and review and 
> discussion of patches can show up other things that need coverage as well 
> (e.g. the extra coverage of strtod not changing errno when it's not meant 
> to arising from discussion of the patch adding coverage for errno setting 
> on overflow).  That's not very suited to posting or reviewing together at 
> present.

I agree with your position, and it would be nice to have dependencies expressed for patches
in patchwork e.g. "Depends on Message-Id: <>" in order for a tester to be able to walk the
patches backwards, such support doesn't exist today.

How do we make it easier for a reviewer to get and test the entire loose collection?

Can we write some script that uses dependency information encoded in the commit message?

It looks like we could use b4 0.14+ features for this?
https://b4.docs.kernel.org/en/latest/contributor/prep.html#working-with-series-dependencies

-- 
Cheers,
Carlos.