Hello, We've just started trialing Hydra at work and are very impressed with the results - it's a lovely bit of kit! I feel we'd be able to use it even better if there was a way to have what I'm calling "all or nothing" channels.
We have a jobset that consists of the following jobs: * frontend - builds the website, minifies scripts, etc * metronome - compiles a Haskell web server * fynder - compiles a Haskell library used by the web server * acceptance-tests - runs end-to-end acceptance tests for the frontend against metronome. I'd like to create a channel that contains the `frontend`, `metronome` and `fynder` packages. The jobset channel gives me exactly this, but unfortunately the channel is updated too frequently. The jobset channel updates when *any* job successful builds, irrespective of the other jobs. I've had a look into `releaseTools.aggregate`, which certainly seems to be going in the right direction. Now we have a `fynder-stack` job that has the above 4 jobs as constituents. I was hoping that I could use this as a channel, but unfortunately this doesn't have anything as build contents. I suppose what I'm trying to do is a small scale version of the nixos channels themselves. If I understand correctly, these only get updated if the `tested` job (which is an aggregate job) passes. Would it be possible to do this entirely in Hydra without scripting? If it requires scripting, could anyone show me what "listens" for success of the `tested` job on Hydra, so I can try and copy the approach? Thanks to everyone working on Hydra, looking forward to making it a key part of our infrastructure! - ocharles
_______________________________________________ nix-dev mailing list nix-dev@lists.science.uu.nl http://lists.science.uu.nl/mailman/listinfo/nix-dev
