It's not a problem to require bash to run the script, since it will be installed inside the autopkgtest testbed due to the package's priority.
I really think we should use pipefail here because of two things: first, you're not explicitly checking the return status of each command (due to "set -e", which is problematic by itself, but let's not go there), and second because of how you're calculating the value of ${content_type}:
Without pipefail, this will always succeed. Even though you're explicitly checking the content of the variable below, it's nevertheless a good practice to fail early here.
I hope this makes sense, but let me know if you still have reservations about using bash explicitly.
Thanks, Mitchell.
It's not a problem to require bash to run the script, since it will be installed inside the autopkgtest testbed due to the package's priority.
I really think we should use pipefail here because of two things: first, you're not explicitly checking the return status of each command (due to "set -e", which is problematic by itself, but let's not go there), and second because of how you're calculating the value of ${content_type}:
content_type=$(curl -s -I http:// localhost/ $ubuntu_ logo_path | grep Content-Type | cut -d ' ' -f 2- | tr -d '[:space:]')
Without pipefail, this will always succeed. Even though you're explicitly checking the content of the variable below, it's nevertheless a good practice to fail early here.
I hope this makes sense, but let me know if you still have reservations about using bash explicitly.