[Dwarf-Discuss] constants updates, minor DWARF standard upgrade?

[Michael Eager]

On 07/20/2012 03:14 PM, Robinson, Paul wrote:
> Michael Eager wrote:
>> On 07/20/2012 12:56 PM, Frank Ch. Eigler wrote:
>>> [...]
>>> Would the committee consider a faster-track number-assignment process,
>>> kind of like IANA, for proposals that do not constitute material
>>> structural changes to DWARF, but mere non-conflicting assignment of
>>> magic numbers?
>>
>> We don't want to have competing and conflicting standards procedures.
>> We have a simple low-overhead standards process.  Adding a fast-track
>> process would increase the complexity and increase overhead (i.e., my
>> time), and might lead to the awkward situation where the fast-track
>> consideration of a proposal and a more deliberative one disagree.
>
> I agree with Michael...
>
> ...except for the one specific case of language codes.
>
> Language codes are a list of values that don't affect anything anywhere else
> in the spec, with the single trivial exception of the default lower bound for
> arrays, which is already incorporated into the language-code table.
> I flipped through Chapter 7 and it's the only case like that, a list of numbers
> that have no other syntactic implications and (almost) no semantic implications.
>
> New language codes are rare, maybe 1 per year. No new language code
> request has been rejected, according to the resolved-issues lists.
>
> The interval between publishing DWARF 3 and DWARF 4 was 5 years.
> I don't see any particular value in waiting 5 years to formally and officially
> assign a name/number combination for a language code.

As said earlier, compiler developers can make their own choices
about whether to wait or to assume that there will be no changes.
With language codes, there's little risk in this assumption.

-- 
Michael Eager	 eager at eagercon.com
1960 Park Blvd., Palo Alto, CA 94306  650-325-8077