hi Cosimo; On 15 January 2015 at 09:58, Cosimo Cecchi <[email protected]> wrote:
> It's of course better to start small and then expand rather than the > opposite, but without getting into the specifics too much, the example of > Telepathy is something I would include in the platform, since from the > client perspective it's convenient to have one contact store managed by the > OS and one defined API to access it; it also makes it possible to have a > "contacts" kind of permission in the sandbox. Telepathy as an API to ship in the SDK is mostly fine, though it still depends on dbus-glib which means additional terrible dependencies in the SDK. we should really start prototyping the actual contacts "portal" API, though. > With regards to python, I tend to agree with you on where the line should be > drawn, even though I am not an expert on the details of its standard > library; in general I think it would be an interesting exercise to put > together a minimal build with whatever is necessary to get pygobject > running. another argument to keep the Python dependency list small is that for every piece of software we include in the GNOME platform we get to be responsible for security updates. we can easily control what we create, and rely on a bunch of other projects for lower levels of the stack, but taking on the whole of Python's standard library may require far too many resources, unless we choose a suitably well-maintained, slow-moving subset, which would imply python2 instead of python3. ciao, Emmanuele. -- https://www.bassi.io [@] ebassi [@gmail.com] _______________________________________________ gnome-os-list mailing list [email protected] https://mail.gnome.org/mailman/listinfo/gnome-os-list
