[PATCH 4/4] pcapng: move timestamp calculation into pdump

[Kevin Traynor]

On 06/10/2023 10:10, David Marchand wrote:
> On Wed, Oct 4, 2023 at 7:13 PM Stephen Hemminger
> <stephen at networkplumber.org> wrote:
>>
>> On Mon, 2 Oct 2023 10:15:25 +0200
>> David Marchand <david.marchand at redhat.com> wrote:
>>
>>>>
>>>
>>> Bugzilla ID: 1291 ?
>>>
>>> This patch (and patch 3) updates some pcapng API, is it worth a RN update?
>>>
>>>> Fixes: c882eb544842 ("pcapng: fix timestamp wrapping in output files")
>>>
>>> Is it worth backporting?
>>> I would say no, as some API update was needed to fix the issue.
>>> But on the other hand, this is an experimental API, so I prefer to ask.
>>>
>>>
>>>> Signed-off-by: Stephen Hemminger <stephen at networkplumber.org>
>>
>> Good question.
>> Is experimental API allowed to change in a stable release?
> 
> I don't think this is cleary described in our ABI policy.
> An experimental API may be changed at any time, but nothing is said
> wrt backports.
> 
> Breaking an API is always a pain, and for a LTS release it would
> probably be badly accepted by users.
> 

yes, I agree. IIRC, this arose sometime in the past with a branch that 
Luca was maintaining and I think the consensus among LTS maintainers was 
not to change experimental API on stable branches.

> Cc: Kevin for his opinion.
> 
> We may need a clarification on this topic in the doc.
> 
> 

Perhaps it's not a "rule" since experimental API comes with no 
guarantee, but I can add something to the docs that it is a guideline 
not to break experimental API on stable branch.