@paulblum ... great to see a new example! The current CI failures should be relatively easy to fix by specifying thresholds (just had to fix a similar thing for #927 - I admittedly didn't investigate and just copied values).

Beyond, I think it may make sense to switch the mechanism from gri30.yaml to h2o2.yaml here and for custom.py (other than not having N2 - which is replaceable by AR for the purpose of an example - this shouldn't change much).

@ischoegl Any reason to prefer h2o2.yaml over gri30.yaml? I doubt there's any difference significant difference in how long they take to run on modern processors.

@bryanwweber ... it's admittedly a minor (philosophical?) point, but things do add up. I would generally advocate for using the smallest possible mechanism that will do the job especially when there is absolutely no benefit of using a larger mechanism: as h2o2.yaml is available (as GRI minus carbon/nitrogen), it's the logical candidate for probing a H2-O2 system for the sake of an example.

PS: As another benefit, end-users (who are probably interested in some C species) will have to actually think about what mechanism to use instead of always defaulting to gri30.yaml - I believe we all agree that there are better options out there.

Just updated the regression test tolerances from the defaults to threshold=1e-10 and tolerance=2e-4, which match the ones specified for cxx-combustor.

@ischoegl thanks for the feedback! I'll look into switching this mechanism next.

@paulblum @ischoegl I'd be in favor of deferring the switch to h2o2.yaml to a separate pull request, since it could be done for several examples, and leaving this with gri30.yaml means it'll match the Python example until such time as we switch any other examples. Doing a PR to change over the example input files would be a good first issue for any new contributors as well.

🤷‍♂️ ... makes no difference as long as it’ll get done. I agree that this is an easy fix ... 5ish lines spread over a couple of files? Not sure it’s worth its own PR ...

PS: #932

Thanks for the review @speth, these were all really good suggestions! Just pushed a commit with the updates, please let me know if there's anything else you think can be improved 😄

Failed CI test due to regression test tolerances, I had fixed that in an earlier version but looks like it got removed by a conflict... latest commit should be all set!

Thanks, @paulblum, this looks great. I think the only thing I'd like to do before merging is to squash the last couple of commits to avoid the large diff associated with the changes to the custom_cxx_blessed.csv file. If you're comfortable doing that, please go ahead, otherwise I can do it.

@speth that makes sense to me, although I'm not familiar with how to squash commits... probably better for you to do it if it's not too much trouble. Thanks again for the help!