On 25 Jun 2015, at 16:59, Ian Campbell <ian.campb...@citrix.com> wrote:
>
> On Wed, 2015-06-24 at 16:08 +0100, Ian Jackson wrote:
>> Rob Hoes writes ("[PATCH RFC 2/9] libxl idl: allow implicit enum values"):
>>> Introducing two special enum values:
>>> * ENUM_NEXT: equal to the previous value in the enum plus 1
>>> * ENUM_PREV: equal to the previous value in the enum minus 1
>>>
>>> This makes it a little easier to maintain enums for which we do not care too
>>> much about the exact enum values.
>>
>> It means that enum values can't be looked up without compiling the
>> code. It also means that deleting an old enum value would result in
>> unexpected and undesirable changes to subsequent enums.
>>
>> So I'm afraid I don't agree with this approach.
>
> Neither do I.
>
> An unstable ABI is one thing, having error messages change randomly from
> "error -3" to "error -4" because someone added a new one is a step too
> far though I think.
>
Fair enough, I thought I’d just give it a go and see what happens. :)
My other option was to leave some gaps between error codes so that you can
easily fit some in between in a logical order. Anyway, I am first going to
focus on finding patterns and possible groupings of error codes without caring
too much about the numbering.
Rob
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
