[PATCH v2] eng: Requirements counting shall start at zero

[Chris Johns]

On 14/12/20 6:20 pm, Sebastian Huber wrote:
> On 14/12/2020 07:31, Sebastian Huber wrote:
> 
>> On 11/12/2020 17:43, Joel Sherrill wrote:
>>
>>>     If renaming them introduces challenges, then we should explicitly
>>>     use -0 for the first requirement always, and not allow unnumbered
>>>     ones to exist.
>>>
>>> +1
>>>
>>> I would be concerned that the unnumbered ones would be following a
>>> different naming pattern which would make the requirements file matching
>>> pattern harder.
>> We should not make things more complicated than necessary. The goal is to have
>> descriptive names for the requirements. If a numbering makes sense, then it
>> should start with zero. If you need more than 10 numbers, then using 00 is
>> also acceptable, for example: /rtems/event/if/event-00.
> 
> It seems the *-0 scheme was already used by Frank to specify the basedefs:
> 
> https://git.rtems.org/rtems-central/tree/spec/rtems/basedefs/req
> 
> For example
> 
> https://git.rtems.org/rtems-central/tree/spec/rtems/basedefs/req/packed-0.yml
> 
> https://git.rtems.org/rtems-central/tree/spec/rtems/basedefs/req/packed-1.yml
> 
> https://git.rtems.org/rtems-central/tree/spec/rtems/basedefs/req/packed-2.yml
> 
> I would have probably used no numbers and instead something like
> 
> packed-member
> 
> packed-type
> 
> packed-enum

Agreed. Is it too late to change?

> I don't care much which style we use in the end, but it should be consistent
> across the specification. So, it is good to have this discussion and it is
> evident what the current

I agree. It is hard to move, remove and rename. It is easy to add.

Chris