Hi Abderrahim,

On Thu, 2022-12-01 at 09:24 +0100, Abderrahim Kitouni wrote:
> Hi Tristan,
> 
> Le ven. 11 nov. 2022 à 09:23, Tristan Van Berkom
> <[email protected]> a écrit :
> > apache-buildstream-1.95.5.dev0.tar.gz
> > -------------------------------------
> > sha256:
> > e48db435eee96b317b1a44d2ce6f7901160e763426b443e1e3461b562daf7863
> > sha512:
> > abbe4b673c28ab103c0f3c93cd61216a54ce73dbb9b1e721fb3a1a13501dde30fd5
> > f19cf2a7bb46ca489590a25f5d09dd9adc76f7889348bd3716102dd0fae4e
> 
> This is mostly ok. I just noticed something in the rat output, the
> following files are missing a license header
>   BuildStream-1.95.5.dev0/src/buildstream/_testing/_utils/junction.py
>   BuildStream-1.95.5.dev0/src/buildstream/_testing/_utils/site.py
>   BuildStream-1.95.5.dev0/src/buildstream/_testing/_yaml.py
> 

These are good catches, I may have overlooked them thinking they are
exempted as being a part of test data, but they are in fact test code.

> Also it's hard to go through the output of rat because of the "false
> positives". We should probably have an ignore file.

My hope is get through this and responsibly track what happens from
now on, I have no expectation for the RAT tool's overly verbose output
to ever be useful in the future.

If there is a contributor that is interested in making the rat tool
produce useful output for our project, they are certainly welcome to do
so.

> > apache-buildstream-plugins-1.95.5.tar.gz
> > -----------------------------------------
> > sha256:
> > 59864f8b7860a01c3c91a61febb59f53ec952644a49243cedfd93d72e0f9d8f2
> > sha512:
> > f621b932d7f52bc351854a4892aa31dc0b64314475ad2c1b5404098c52686bc04f4
> > 8e01334353f34f58207879dcf2e3c20b09b97d4c13c61a04704009822b31c
> 
> This one looks good. I would just like to discuss the cache key
> stability promise, and what can be done about things like
> https://github.com/apache/buildstream-plugins/issues/34  and
> https://github.com/apache/buildstream-plugins/pull/38 if we release
> 2.0 without them.

I've noticed these, and my inclination is to just not apply them - but
if we're going to go for yet another round of 2.0 candidates we might
as well apply them.

After that, cache key changing commits will be forever closed,
regardless of how the default configurations for these plugins may be
considered "imperfect", that imperfection is rather unimportant.

Cheers,
    -Tristan


Reply via email to