python-nss feature
John Poelstra
poelstra at redhat.com
Fri Aug 8 21:00:36 UTC 2008
John Dennis said the following on 08/08/2008 01:10 PM Pacific Time:
> Paul Fields graciously has brought to my attention some concerns
> expressed here as to whether python-nss is truly a feature or just mere
> packaging. The feature page was probably deficient in explaining why
> this is a feature. Previously there were no python bindings for NSS. NSS
> is our preferred cryptographic library for SSL/TLS and certificate
> management (largely because of it's FIP 140 certification, plus a
> variety of other issues). Python is widely used in Fedora. The absence
> of Python bindings for NSS has been a developmental liability for many
> Fedora projects. Due to the complexity of NSS and the desire to produce
> a binding which was "pythonic" it was not possible to produce a binding
> via automated tools (e.g. swig). Instead the binding was written by hand
> with Red Hat investing many man months of engineering in the effort. If
> the binding had been easy to produce one would have existed already, but
> it's been missing for years. This is why it's a feature. It's not "just
> packaging" because that implies the binding already existed and we just
> packaged it up, something which would have been just a fraction of the
> effort actually invested. To my mind a feature is something which did
> not exist previously and brings significant new functionality to the
> release, I believe python-nss fulfils that criteria.
>
> I will update the features page with this information.
>
+1
Considering that this functionality is originating in Fedora I also
think it should be considered a feature.
I think it meets at least two of the points stated here:
https://fedoraproject.org/wiki/Features/Policy/Definitions#Definition_of_a_Feature
I don't think it applies to the situation above, but really, what harm
is there in erring on the side of calling something a feature when maybe
some people may consider it isn't? We've got a nice list of features
firming up for Fedora 10, but surely there is room for many more on this
page: https://fedoraproject.org/wiki/Releases/10/FeatureList
I'd rather we take a harder line on not calling things features when we
get to a place when we have too many features. I don't think we are
there yet :)
John
More information about the devel
mailing list