Add macro do_drat()
#221
Add macro do_drat()
#221
Conversation
| remote_url = NULL, | ||
| commit_message = NULL, | ||
| commit_paths = ".", | ||
| ssh_key_name = "id_rsa", |
There was a problem hiding this comment.
Is this still the correct default?
There was a problem hiding this comment.
In this case yes. We have to manually insert the private secured key in appveyor.yml and I think inserting it with "TRAVIS_DEPLOY_KEY" would be misleading.
Since "id_rsa" is supported universally, I choose to go with it here.
Keys on Appveyor needs to be manually created anyway until now. We could change the name once we have an Appveyor client pkg.
There was a problem hiding this comment.
I wouldn't mind reverting to "id_rsa", this would save lots of downstream pain.
There was a problem hiding this comment.
Hm, I know. One reason of the whole effort was to ensure that differently named private SSH key works throughout all functions, especially when installing the key. This came from the motivation to decouple travis::use_travis_deploy() from {tic}. I know that this does not mean that the default needs to change necessarily - but having a different one is an indirect way to save the tests whether a custom name would also work 😅
Also I still think that some non-tech users will have an easier life understanding what that env var is about.
However, I see that we complicate things (a bit) since id_rsa is the canonical default picked up everywhere.
Argh, I don't know what's best 🤔
Uff, this took some time but I really like it. Hence I'd say it was worth the time 😃 |
Because I have a use case for it, I optimized the drat approach and put it into a macro.
Changes
deploy_dev = TRUE.do_drat()(more descriptive, less complicated compared to the old approach)index.Rmdto have a file listing. I think we can do better and easier within README.Rmd. The rendering of README.Rmd should happen externally and not somewhat hidden within that macroI am about to update the tic.drat example repo as well but I am still facing some deploy issues on Appveyor which might relate to ropensci/ropenscilabs account permissions. With the linked example above for the {mlr3learners] repo everything works smoothly :)