Messing around with .travis.yml

[ewels]

I think that the builds are still timing out for some reason.

Also want to check that nf-core doesn't run with --release as this PR goes to the dev branch.

[ewels]

Dockerhub is making me sad 😞

[ewels]

Oops, accidentally opened a duplicate PR #2

---

Tests failing as I can't create the repository on dockerhub currently.

• Remove groovy syntax strings
• Playing with tidier installation script and travis config
• Minor pipeline name tweaks

---

Now the tests are failing because the new nfcore/base image isn't ready. Hopefully it will be soon and the tests can be rerun.

[maxulysse]

I can see that you simplified the actual Travis tests, but that way, you'll have to specify the Singularity version here, and not in the tests.
Wouldn't that be more complicated in the end?

[ewels]

I figured that it's pretty rare to need to update the singularity version.. How often does that change?

[ewels]

(in contrast to the Nextflow version, which changes a lot)

[ewels]

Hmm, yeah ok - fairly often I guess... http://singularity.lbl.gov/all-releases

So in this case should we also have a "minimum singularity version" then?

[ewels]

HiSAT2 tests now passing 🎉

STAR tests still timing out as they always have. Previously we just didn't run these tests, but that feels kind of unsatisfactory.

[maxulysse]

Have you tried with travis_wait 50 or something like that?
cf SciLifeLab/Sarek@bc408e0

[ewels]

I tried travis_wait 30 and it failed, but I think that may have been for a different reason. Maybe I'll try again..

[maxulysse]

I noticed that afterwards...
I never managed to make the travis_retry work, but the travis_wait was working well until I managed to simplify our tests.

[ewels]

I'm going to force a merge here - the code is fine, it's just the travis tests which are being stupid and they can be fixed in a future commit.