feat: allow users to run custom commands before exit #3
Conversation
|
|
||
| finish () { | ||
| exit_code=$?; | ||
| if [ ! -z "${CALLBACK_ON_SUCCESS:-}" ] && [ "$exit_code" = 0 ]; then |
There was a problem hiding this comment.
I'm a bit torn on whether this should be a single CALLBACK_COMMAND that gets passed the exit code as an argument, but considering this gets passed as an env var, it's probably a bit annoying to start adding conditions to that single line string...
There was a problem hiding this comment.
yeah, I can totally see that; probably easiest to do this
| finish () { | ||
| exit_code=$?; | ||
| if [ ! -z "${CALLBACK_ON_SUCCESS:-}" ] && [ "$exit_code" = 0 ]; then | ||
| sh -c "$CALLBACK_ON_SUCCESS" |
There was a problem hiding this comment.
I wondered if we should worry at all about cleverly escaping this then reminded myself that really we don't want to worry since this is basically arbitrary remote code execution by design anyway.
Maybe if we were worried about this we would instead provide a webhook type URL rather than totally arbitrary command
| @@ -36,6 +37,18 @@ The OAuth access token used for authenticating against the target wiki. | |||
|
|
|||
| The OAuth access secret used for authenticating against the target wiki. | |||
|
|
|||
| ## Running commands after the script has finished | |||
There was a problem hiding this comment.
prasie: thanks for diligently adding the docs here; that's super nice
Ticket https://phabricator.wikimedia.org/T369261