The 'wait-for' command line utility

Sep 8, 2016

As a software developer, quite often there are things you cannot do until some semi-long process has completed. For example, waiting for a new build, an automated test run or when deploying to a cluster and you want to know when all machines have finished updating.

If you’re running a local build you would do build ; show-notification 'build done' (e.g. on Debian/Ubuntu you’d use notify-send for this). However, if you’re waiting for a build on a CI machine you typically have to reload some webpage to see if it’s done or query job progress by curling some API.

To be able to get a notification when such jobs are done, I wrote a Python script called wait-for that repeatedly runs some command until the command prints some particular string (or returns a specified exit code, or prints something different from what it printed last time it was invoked).

This script makes it easy to do stuff like for example:

$ # Suppose OpenSSL (or similar) has announced that they will publish a new
$ # security fix today and you want a notification when they publish the details.
$ CHANGELOG="https://www.openssl.org/news/changelog.txt"
$ wait-for --stdout-change "curl -s $CHANGELOG" ; notify-send "It's published"

$ # Suppose you have a scruffy old Jenkins server, and you want to wait for the
$ # next successful build and then deploy that to some staging environment.
$ URL="http://jenkins.ecorp.com/job/SOME_JOB_NAME/lastSuccessfulBuild/api/json"
$ wait-for --stdout-change "curl -s $URL | jq .number" ; deploy-latest-successful

The wait-for source code is available here; it even has a few tests. The tests were written in bash because that was obviously a good idea (my condolences to the kittens that died as a result).

EDIT: After posting this, someone pointed out the fact that watch --chgexit cmd covers the above use case, so wait-for is only useful if you actually need --stdout-contains or --stdout-equals etc.

• tools