Add support for processors facts on *BSD #489
Conversation
|
Can one of the admins verify this patch? |
smortex
force-pushed
the
freebsd-processors
branch
from
c1d0370
to
cda57f5
Compare
|
|
smortex
force-pushed
the
freebsd-processors
branch
from
c508b30
to
3fad4b8
Compare
|
|
||
| module Facter | ||
| module Freebsd | ||
| module FfiHelper |
There was a problem hiding this comment.
Please add some tests for Freebsd :: FfiHelper
There was a problem hiding this comment.
Did it, but it raise a few concerns for me because these tests just mock everything and we are testing the mock framework more than anything else 😢
|
I realized that most of the code could work on all *BSD systems with just a few tweaks, so I refactored the PR a bit to move the code from the FreeBSD to the BSD directories (Commit "(maint) Make the code more BSD-generic"). Yet, some differences exist for some aspects of the processors fact, and I could not manage yet to do things correctly. The BSD and FreeBSD resolvers are almost the same, only the way of fetching the CPU speed is different, and I would like to avoid duplicating all the code in this class. I though about inheritance, but it is not really working because both classes are instanced and the common facts are fetched twice :-/ |
smortex
force-pushed
the
freebsd-processors
branch
5 times, most recently
from
111acf4
to
7a1039d
Compare
smortex
changed the title
smortex
force-pushed
the
freebsd-processors
branch
from
7a1039d
to
9bc4659
Compare
|
We had some problems with the GitHub CI, but now it's fixed. So a merge with master will fix it for your PR as well. |
smortex
force-pushed
the
freebsd-processors
branch
2 times, most recently
from
8398521
to
fb88cc5
Compare
smortex
force-pushed
the
freebsd-processors
branch
3 times, most recently
from
b1e7f1b
to
c81ad64
Compare
|
Inheritance in resolvers not being really helpful, I completely separated the Bsd resolver and the Freebsd one. That way, facts are not collected twice by the Bsd and the Freebsd resolvers on FreeBSD as it was previously the case. I reabased the work on top of master, and since the commit where going back and forth, I squashed them to make the change more readable. Please review, thanks! |
|
@smortex please rebase on master to fix RuboCop. Regarding |
smortex
force-pushed
the
freebsd-processors
branch
from
c81ad64
to
1199ec3
Compare
For BSD platforms, rely on a sysctl(3) FFI wrapper for getting processors information. On FreeBSD, rely on sysctlbyname(3) FFI wrapper. This allows Facter 4 to return the same processors facts as returned by Facter 3 on FreeBSD.
smortex
force-pushed
the
freebsd-processors
branch
from
1199ec3
to
1c91d4b
Compare
|
@IrimieBogdan done! |
Add a basic sysctl(3) FFI wrapper for getting processors information. This allows Facter 4 to return the same processors facts as returned by Facter 3 on FreeBSD.
While working on this I spotted a typo in an unrelated file, so fixed it in another commit.