Bart Smaalders wrote: > Shawn Walker wrote: >> Nicolas Williams wrote: >>> As for Shawn's point about headers, I'd rather just always install them, >>> rather than conditionally install them. Non-developers won't notice, >>> and minimizers can always filter out any headers. >> >> I disagree. There are multiple benefits that all help different >> groups of people: >> >> * Sun (savings on bandwidth, etc.) >> >> * Metered-Bandwidth Users (e.g. most of Australia, Europe, etc.) >> >> * Embedded and resource-constrained Users >> >> ...others. >> >> It's silly to install headers if you don't have any software that can >> use them. >> >> Cheers, > > Since both variants and facets are defined by the publisher > rather than the client, publication and signing of manifests that > contain such tags indicates acceptance/support of those variants. Ad-hoc > client side filtering allows arbitrary removal of content from packages > and thus violates support and signature boundaries. > > As a result, I think filtering should be an exceptional mechanism, not > standard operating procedure; removal of header files should not require > rendering system unsupportable.
I'm a little confused here. First, I thought we were planning on ad-hoc client side filtering anyway that was directly under the user's control. Second, I was advocating for the variants and facets by the publisher to be used here for the selection or installation of headers. Maybe I don't have a great understanding of how facets, variants, and filters are all supposed to fit together. I personally believe there is an opportunity here to make the system more efficient by using this additional information to drive installation behaviour. That's all. Cheers, -- Shawn Walker _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
