Bye Bye Travis CI
daniel.haxx.se
daniel.haxx.se
I assume most projects on github have switched to github actions.
What I wonder though is: how do they detect and block crypto mining?
Edit: https://github.blog/2021-04-22-github-actions-update-helping...
And that's the whole point of open source. Free to use as one wishes.
At some point, our `master` branch broke (unrelated to CI), so I picked up fixing that. We ran out of credits mid-way, so work on the entire project was blocked.
We requested more credits and had no news. I eventually moved the project over to using sourcehut's CI (although keeping the source on GitHub).
Weeks later, we got the credits from Travis. It was too little too late.
Travis I've heard not so great things about since the buyout. GitHub actions is great (also one less vendor is a benefit) but Circle were worth the money, mainly because of better docs and sales engineers who really wanted to fix out problem. We had a monolithic app that sometimes took 40 mins on a (badly maintained) Concourse setup that changed to about 2m40s on Circle. I bet we could have done the same on GHA but it would have taken longer.
If you are being paid to make a free product, it's actually quite reasonable for the developers of tools you use to be paid too. I see absolutely no issues, ethical or otherwise, with this.
The only iff-ish bit is Travis' legacy committment to serve open source projects for free, but again I suspect that some of the larger ones ultimately abused the offer and that led to the change.
> If you are being paid to make a free product, it's actually quite reasonable for the developers of tools you use to be paid too. I see absolutely no issues, ethical or otherwise, with this.
I don't think Travis was called unreasonable anywhere. The post even ends with "Thank you Travis CI for all those years of excellent service!", which I would describe as thankful, not dissatisfied. Also the "Services come and go. There’s no need to get hung up on that fact but instead keep moving forward with our eyes fixed on the horizon.".
You don't see the donut hole issue? "Sorry, but if you want to sponsor the project you need to cover 100% of the cost. Don't worry, I'm sure the director of your department will be really receptive to the suggestion and not view you as a potential problem." It would effectively result in corporations buying a select few projects, which wouldn't end well for anyone... imagine patent portfolios but with more drama and a legal framework closer to what you'd find in Bartertown from Mad Max.
So what happens when more than one of these "developers of tools you use to be paid" demands the same? You get Mark Zuckerberg shouting "Who run Bartertown?!" at some poor project maintainer.
Edit: It just occurred to me that the point of confusion might be my reference to the 90s era US welfare program, and not Master Blaster... basically the donut hole described the gulf between how much somebody on welfare got paid while not working and how much they'd need to get paid in order to cover the loss of access to the program (because having a job, among other things that help to lift one out of poverty, made you ineligible). The really bad part has more to do with the how unlikely it was for somebody in such a situation to ever be able to demand such a rate without first developing a work history and gaining experience - which guaranteed no positive change without first assuming a massive amount of risk. The same phenomenon would occur here, where taking a single dollar or accepting a pull request from an address with a corporate domain would have to be weighed against the cost of all tooling in use by a project suddenly demanding money. It would be a massive disincentive to incremental improvement, where the only rational move is to either abandon all such tooling (which, ironically, very likely depends on other open source projects in a similar situation) or turn down every offer that doesn't fully cover all potential costs.