ACK

ACK

On Tuesday, October 6, 2015, Wladimir J. van der Laan <
[email protected]> wrote:

ACK

—
Reply to this email directly or view it on GitHub
#6762 (comment).

Next time we should try to look for these things in the review, so they don't require lots of additional pulls to fix up.

Yup. Maybe we should add the result of git status to the travis build logs?

I think that's a good idea @theuni can we add a 'tree is clean post-build' check to Travis? It may be complicated by the fact that we're building from a make distdir.

It's ugly, but this works locally:

diff --git a/.travis.yml b/.travis.yml
index c878514..68362f1 100644
--- a/.travis.yml
+++ b/.travis.yml
@@ -70,4 +70,5 @@ script:
     - if [ "$RUN_TESTS" = "true" ]; then make check; fi
     - if [ "$RUN_TESTS" = "true" ]; then qa/pull-tester/rpc-tests.sh; fi
 after_script:
-    - if [ "$TRAVIS_PULL_REQUEST" != "false" ]; then (echo "Upload goes here. Something like: scp -r $BASE_OUTDIR server" || echo "upload failed"); fi
+    - echo "Checking for left-over files..."
+    - (export GIT_WORK_TREE=.; git ls-files "*.gitignore" | git checkout-index -q --stdin && git ls-files --error-unmatch -- `git ls-files --exclude-standard -o`) && echo "OK"

It grabs the missing .gitignore files (they're not distributed), then uses a GIT_WORK_TREE trick to fool git into acting on the distdir.

It's rather kludgy for the travis file, but I can't imagine it's useful enough to break out on its own. Want me to PR it that way?