Build Bandits - who broke your build


(Richard Gubby) #1

Hi all,

Thought you might like to hear about a side project i’ve been working on for a while.

It’s called Build Bandits and it’s for tracking all of your CI builds and reporting who broke them in a fun kind of way (as western style bandits). It originated as a hackday project at work, but I took it a bit further and made it into it’s own service that anyone could use.

You can mix and match the CI services - so if you have some builds in Travis, some in Circle, and so on - then you can still track them all. The committer with the most broken builds has the biggest reward!

Here is the important bit - the URL: https://buildbandits.com

And here are a few that I prepared earlier:


Any feedback and comments would be very much appreciated!


(Daniel Hollands) #2

I’d love to give this a try, but we use Solano Labs for our CI (and we’re only using them because Semaphore CI doesn’t support GitLab)


(Marc Cooper) #3

Out of interest, what are you missing from GitLab’s CI that results in you using Solano?


(Daniel Hollands) #4

The knowledge of how it works. I think we’d be more than happy to use (our self-hosted) GitLab for this, but we’ve not had the time to sit down and figure out how to provision the CI server (the runner, is it called?) works. I’d welcome any help and insight into this.


(Marc Cooper) #5

I’m in a similar position to yourself. I run simple CI on GitLab – click the button for the default script then edit. e.g.


It works well. I haven’t built anything substantial on it. I’ve been scanning the updates about it and have been wondering if it’s time to spend the effort to learn it.


(Jon) #6

Bloomin’ 'eck! I thought I’d stumbled on a marvellous deal when I found I could use 1,500 free minutes per month on CircleCI. Now I discover, had I gone for GitLab, I’d get 2,000 free minutes per month.

Well, ho hum, will add GitLab to the hand-wavy future investigation list :smile_cat:


(Andy Wootton) #7

I have mixed feelings. I’ve worked in a “no blame culture” and that was a licence for gun-slingers to shoot other people in the foot but I’m suspicious of metrics that might cause wrong behaviors too. In agile (all dev?) environments the team needs to be co-operative rather than competitive. You don’t want people to be discouraged from checking in regularly, addressing the more difficult bugs or just doing more coding than other people, in case they break the build and top the chart. I think I’d collect the information and have debrief sessions about broken builds, so everyone learns from them and people are only embarrassed when they repeatedly don’t learn from their mistakes. We don’t all suffer equally from the same level of public humiliation, so it isn’t a very fair punishment.

Seems relevant: I almost wrote an essay on how wrong this is but it’s also weirdly true of a massively dysfunctional team. Except for the bit where there’s nothing on the Kanban board. https://www.facebook.com/UNILADTech/videos/2021793774765000/


(Marc Cooper) #8

Do you have a non-Facebook version? I have Facebook /dev/null’d in the firewall. I consider it the DarkNet, if you will.

Also, if you have a no-blame culture, and folk are shooting each other – figuratively or otherwise – then, by definition, it’s not no-blame. And it’s not a metric either. But yes, collaborative trumps competitive in s/w dev – and in most other things, imo.

I agree strongly with your comment about public humiliation. Well put.


(Andy Wootton) #9

No, I think it’s actually hosted on FB. It’s just a micky-take of streotypes but it’s weirdly wrong and I’m not sure if it’s deliberate or they just didn’t understand what they observed. It’s well filmed.

It’s on Twitter too https://twitter.com/uniladtech/status/910429822379286529?lang=en


(Jon) #10

Heh, agreed - it’s the spawn of the devil. It’s also here.