Delegate to unified strip-marker-sections script

[jotaen4tinypilot]

Resolves https://github.com/tiny-pilot/tinypilot/issues/1710. Stacked on https://github.com/tiny-pilot/tinypilot/pull/1720.

Based on the simplifications in unset-static-ip, this PR removes the code redundancies in the unset-static-ip and change-hostname privileged scripts, and calls the new, unified strip-marker-sections script instead.

Notes

• change-hostname:
  • Naively speaking, it would seem simpler to append the new entry to the file. I remembered, however, that the prepending was done on purpose, because in /etc/hosts the first matching entry takes precedence. I captured that as a comment therefore.
  • I slightly reordered the code so that all /etc/hosts-related logic is close together now (i.e., clearing the markers, then rewriting the file).
• unset-static-ip
  • ~I couldn’t figure out why we have been using tee previously to write to the config file, because we discarded tee’s stdout output by redirecting to /dev/null. I’m also not sure why we used sudo for invoking tee – we already execute the script with root privileges. It’s a bit odd, because I cannot imagine we did that without good reason~ 🤔 (We had been using sudo tee to avoid permission issues when writing the file.)

[mtlynch]

I couldn’t figure out why we have been using tee previously to write to the config file, because we discarded tee’s stdout output by redirecting to /dev/null. I’m also not sure why we used sudo for invoking tee – we already execute the script with root privileges. It’s a bit odd, because I cannot imagine we did that without good reason 🤔

I think we wrote that assuming the whole script wouldn't necessarily run with root privileges.

But the pattern of sudo tee is one we use in a few places. It's because if we do:

sudo some-cmd > /etc/dhcpcd.conf

Then the command fails if /etc/dhcpcd.conf is owned by root and is not world writable. But if we do:

sudo some-cmd | sudo tee /etc/dhcpcd.conf > /dev/null

It lets us write to /etc/dhcpcd.conf from a root user context.

[jotaen4tinypilot]

Automated comment from CodeApprove ➜

⏳ @jdeanwallace please review this Pull Request

[jotaen4tinypilot]

I think we wrote that assuming the whole script wouldn't necessarily run with root privileges.

Ah, thanks for the explanation, that makes sense then. Do we need to preserve this behavior (in the strip-marker-sections script), or shall we keep the currently proposed > redirect-to-file? For our own purposes (e.g., when calling the scripts from Python), the > works fine.

[mtlynch]

Ah, thanks for the explanation, that makes sense then. Do we need to preserve this behavior (in the strip-marker-sections script), or shall we keep the currently proposed > redirect-to-file? For our own purposes (e.g., when calling the scripts from Python), the > works fine.

No, we can leave those as-is. We only need to do the tee workaround when we're running in a non-root context to begin with so we have to sudo to write to the file.