chmod 600 to config file

[csgillespie]

No description provided.

[eddelbuettel]

Good catch. Not sure why this now failed on Travis CI when it didn't before -- can you poke around?

[csgillespie]

In the second commit I commentted out my changes, i.e. the only difference with the master was a single line comment. It still failed. This implies a travis issue, rather than something to do with my commit.

In your test file you have

stopifnot(file.exists("~/.rpushbullet.json"))

Where/how is this file created?

For info, the test passes on my machine. But I have "~/.rpushbullet.json".

[eddelbuettel]

Where/how is this file created?

In .travis.yml:

rpushbullet/.travis.yml

Lines 19 to 22 in b66c07c

	# decrypt the file .rpushbullet.json, cf https://docs.travis-ci.com/user/encrypting-files/
	# condition this on have secure environment variables which PRs off forks do not
	# cf https://docs.travis-ci.com/user/environment-variables/
	- if [ "$TRAVIS_SECURE_ENV_VARS" == "true" ]; then openssl aes-256-cbc -K $encrypted_988d19a907a0_key -iv $encrypted_988d19a907a0_iv -in .rpushbullet.json.enc -out ~/.rpushbullet.json -d; fi

I think the stopifnot() is your enemy. Maybe just skip tests instead of aborting?

[eddelbuettel]

Also:

rpushbullet/.travis.yml

Lines 11 to 14 in b66c07c

	# condition this on have secure environment variables which PRs off forks do not
	# cf https://docs.travis-ci.com/user/environment-variables/
	- if [ "$TRAVIS_SECURE_ENV_VARS" == "true" ]; then Run_RPushbullet_Tests="yes"; fi
	- if [ "$TRAVIS_SECURE_ENV_VARS" == "true" ]; then Run_RPushbullet_Tests_All="yes"; fi

Seems like you were in fact set up to fail. Sorry about that.

But hey, I generally recommend to people to not open PRs without prior discussion :) [ Mostly for Rcpp et al though ... ]

[csgillespie]

I did know about your PR policy, but I figured a one line PR was just as easy for you to accept/reject as an issue followed by a PR.

I think there's an easy fix for travis

[eddelbuettel]

Do you mind if I just squash-merge it?

[eddelbuettel]

Hm. Now I will only ever get untested PRs. Not sure I like that for the longer term. Ok for now/today.

[csgillespie]

I think the "proper" solution would be for you to add the API key as an encrypted travis environment variable. Then create the json file on travis. A similar idea to the drat/travis stuff

[eddelbuettel]

Could you take the lead on that, "copious spare time" permitting ?