Re: [RFC] renaming packages/ to recipes/

[Koen Kooi <k.kooi@student.utwente.nl>]

On 19-05-09 15:02, Philip Balister wrote:
> Frans Meulenbroeks wrote:
>> 2009/5/18 Rolf Leggewie <no2spam@nospam.arcornews.de>:
>>> Marcin Juszkiewicz wrote:
>>>> My addition to this:
>>>>
>>>> mkdir recipes-broken
>>>> git add recipes-broken
>>>> git mv recipes/obsolete recipes-broken/obsolete
>>>> git mv recipes/nonworking recipes-broken/nonworking
>>>>
>>>> As this will allow users to set:
>>>>
>>>> BBFILES = "/somewhere/oe/recipes/"
>>>>
>>>> which is supported by BitBake for years.
>>> I'd like to revisit this suggestion from Marcin which came up during the
>>> discussion of the packages->recipes renaming. I think the above makes
>>> sense
>>> and should be committed. I think it would be nice if the folder was
>>> named
>>> broken-recipes instead of recipes-broken. First of all, it sounds nicer
>>> IMHO ;-) but more importantly, this works better with bash-completion.
>>>
>>> Reopening discussion for this change.
>>
>> I think adding recipes-broken (or broken-recipes) is a bad idea.
>> It is quite possible that a recipe is broken only for some platforms
>> (e.g. because they use an older libc or gcc, or because functionality
>> is not working in the machine specific parts.
>> In that case I can image we end up in a recipe being shifted around.
>> Not really a good plan.
>> Unless of course we end up with some good rules on when a recipe
>> should be moved to broken.
>>
>> If I look at tinderbox I see various recipes that do not build for
>> some archtectures or configurations.
> A broken recipes directory sounds like a good idea, but I think it will
> not work well in practive. My concern is the history will be filled with
> people miving recipes in and out of the broken directory (for reasons
> mentioned by Frans)
>
> ow about trying to generate a report from tinderbox that lists broken
> recipes?

Sounds like the coverage matrix I proposed at OEDEM 2006 :)

regards,

Koen