https://github.com/osquery/osquery logo
#core
Title
# core
s

seph

02/28/2021, 3:03 AM
Having spent several hours staring at the augeas code, I wonder how many people use it. It felt incomplete. PR up.
This is a somewhat gentle rant. It has not been a fun several hours
It mis-reported the
path
column. It didn’t support
LIKE
on either
node
or
path
.
Etc
t

theopolis

03/01/2021, 5:16 PM
thanks for looking into the table and finding fixes+improvments
do you mind breaking those changes up into distinct PRs?
for example: 1. removing the no-standard-includes searching 2. fixing path reporting 3. implementing LIKE support for node and path
s

seph

03/01/2021, 5:18 PM
Maybe? (1) is a one line change, it’s easy to separate out, but feels silly. OTOH, if it makes discussion easier… (2) and (3) are somewhat distinct code wise, but make more sense together. I guess I can split them if you find it simpler to review
You know what? I’ll split them. I’d rather just split them than talk about it. 🙂 It’s just CI time.
(I’m happy to talk about intent, and the general chaos of that table. But 1 pr vs 3 isn’t really important)
t

theopolis

03/01/2021, 5:26 PM
It's mostly a software dev like-to-have, changes should follow a 'single thesis' rule, which makes them easier to revert and makes issues easier to rootcause and bisect.
s

seph

03/01/2021, 5:28 PM
I understand the arguments in favor, and generally agree with the sentiment. But I have some pragmatism as well. Obviously, we should PR things in small usable pieces, akin to milestones. It’s easy to point at things that are very coarse. And I think I can point at things that are too small. And then there’s a grey area.
Split.
6 Views