On 08/20/12 16:03, Tim Foster wrote:
On 08/21/12 10:12 AM, Brock Pytlik wrote:
On 08/17/12 16:23, Tim Foster wrote:
https://cr.opensolaris.org/action/browse/pkg/timf/pkglint-legacy/pkglint-legacy-webrev/
Thanks for taking a look!
[snip]
pkglint_action.py:
Maybe it's just that I don't really grok what the legacy action is
doing, but this comment doesn't make much sense to me. I think breaking
it up into 4 sentences, one per parameter might make things clearer.
Yes, I'll tidy that up. The pkglint legacy check tries to ensure that
if you have a 'pkg=SUNWfoo' attribute in your legacy action, then if a
user tries to do 'pkg install SUNWfoo', that should result in the
package containing the legacy action being installed (IPS follows
pkg.renamed packages, so pkglint needs to do the same thing in the
course of this check) If SUNWfoo doesn't exist, then we don't care.
This changeset makes us consider packages called SUNWfoo as well as
package-leaf names ..../SUNWfoo as satisfying our rules.
Gotcha. Maybe just putting that in as the description would be enough to
make things clearer. Or maybe I just need coffee today...
A more general question:
With a legacy action, are we guaranteed that the target will be a base
name? Could we (in the future perhaps) have a legacy action whose target
was something like "foo/bar"? In that case, I'm a bit concerned about
line 1022 of engine.py. If that's simply not something we care about
now, then feel free to ignore it :)
Target could certainly be 'foo/bar' eg.
set name=pkg.fmri value=pkg:/foo/[email protected]
legacy pkg=SUNWbar ....
we would expect that if SUNWbar exists (likely renamed) that the
rename-chain includes a dependency on pkg:/foo/bar.
How does line 1022 bother you here? In particular, we're adding that
so that the following package would be an acceptable use of the legacy
action:
set name=pkg.fmri value=pkg:/foo/bar/[email protected]
legacy pkg=SUNWbar
by doing 'pkg install SUNWbar' we get pkg:/foo/bar/SUNWbar installed
(or a message saying it clashes with another package I've not included
the manifest for, pkg:/SUNWbar)
Sorry I should've been clearer. What happens if we have:
legacy pkg=foo/bar
? Like I said, maybe we just don't care for now cause our legacy
packages can never be named like that? I guess I was thinking of what
this might look like if we had other packaging systems as legacy.
[snip]
Thanks,
Brock
cheers,
tim
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
