On 07/27/11 16:09, Brock Pytlik wrote:
On 07/27/11 16:00, Shawn Walker wrote:
On 07/27/11 15:52, Brock Pytlik wrote:
On 07/17/11 21:38, Tim Foster wrote:
Hi there,
I've got an updated webrev here that makes pkglint check the file
actions are using the overlay attribute properly:
https://cr.opensolaris.org/action/browse/pkg/timf/pkglint-overlay-validate-v2/pkglint-overlay-validate-webrev-v2/
...
672: is this assertion saying that I can't allow an overlay on a file
without having at least one package deliver an action which overlays on
it? That seems a bit restrictive to me. If this is a check we do want,
then it seems strange to me that we only check that there is some
variant combination under which an overlay is delivered, instead of all
variant combinations. (For example, if I had a overlay=allow action that
was delivered under all variants, but a overlay=true that was delivered
only on debug, sparc, this check would pass, which seems a bit odd to
me.)
We definitely don't want that restriction. You should be to deliver a
package that overlays a file, or a package that allows overlaying a
file independently. There should be no requirement for either to exist
at the same time.
-Shawn
I do think it's worth checking that, if a package being linted delivers
file /usr/foo/bar with overlay=true and if another known package
delivers /usr/foo/bar, that we check that that package delivers it with
overlay=allow (or that an exclude dependency exists).
Yes, that's likely sane.
-Shawn
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
