cygwin support

[pawelpanasewicz]

Anything against this? Now I can run it as well from windows.

[paulp]

Does travis offer any way to test it?

[xuwei-k]

Maybe appveyor support windows and cygwin.

https://www.appveyor.com/docs/installed-software/#mingw-msys-cygwin

[pawelpanasewicz]

Hi!
Can we test it?

Automated testing after each commit in CI requires some work and decisions.

Firstly - bats must be able to run under cygwin in order to run sbt-extras' tests which will prove that it behaves nicely under cygwin.

There is some work done (in LeahCim/bats fork) towards this direction, but it's not merged so far. LeahCim/bats is developed practically on top of master of original sstephenson/bats (it does not contain changes from particular text file). On the other hand LeahCim/bats is not fully compatible but maybe it's enough to test sbt-extras and it's worth to use it for price of having tested cygwin.
See PR128.

Secondly - it has to be run on agent which supports cygwin. As @xuwei-k notices approyer is one possible choice.

Let me do some research on "firstly" thing.
What is your opinion?

[pawelpanasewicz]

I am almost there.

This is build log from appveyor which runs test under cygwin on windows.
https://ci.appveyor.com/project/pawelpanasewicz/sbt-extras/build/1.0.13

Tests are mainly failing with assertion messages very similar to this one:

# expected:
# Downloading sbt launcher for 0.12.2:
#   From  http://repo.typesafe.com/typesafe/ivy-releases/org.scala-sbt/sbt-launch/0.12.2/sbt-launch.jar
#     To  $ROOT/.sbt/launchers/0.12.2/sbt-launch.jar
# 
# actual:
# Downloading sbt launcher for 0.12.2:
#   From  http://repo.typesafe.com/typesafe/ivy-releases/org.scala-sbt/sbt-launch/0.12.2/sbt-launch.jar
#     To  C:\cygwin64\tmp\download.bats.744\.sbt\launchers\0.12.2\sbt-launch.jar

As one can see only prefix of To some_path are different.

This is because normalize_paths doesn't normalize windows like paths. This is code of normalize_paths:

normalize_paths () {
  stdin_or_args "$@" | \
    sed "s:$TEST_ROOT:\$ROOT:g" | \
    sed "s:$HOME:\$ROOT:g"
}

Windows like path in this example is C:\cygwin64\tmp\download.bats.744\.sbt\launchers\0.12.2\sbt-launch.jar. We need to normalize it into $ROOT/.sbt/launchers/0.12.2/sbt-launch.jar.
How to do this?
The command cygpath -w $TEST_ROOT called from scope of mentioned function can help. It returns wanted prefix of windows style path, in this example it is C:\cygwin64\tmp\download.bats.744.

Now it is sufice to make some replacement in normalize_paths and it should work.

The tricky part for me is that I don't know how to use sed in order to replace this from input string.

Can someone help?

[paulp]

@pawelpanasewicz If I understand correctly having barely skimmed and running out the door, this might do it. Replace sed "s:$TEST_ROOT:\$ROOT:g" with

sed "s:$(cygpath -w "$TEST_ROOT"):\$ROOT:g"

[pawelpanasewicz]

@paulp Thanks! I've manage to run it.

Unfortunately tests are not passing. At first glance outputs are the same. I did some CRLF => LF convertion hoping it will solve the problem but it didn't helped. Not sure what can cause problems. I'll think on it.

https://ci.appveyor.com/project/pawelpanasewicz/sbt-extras/build/1.0.22

Briefly some test failures look like this:

not ok 2 tolerates blank lines and comments in jvm_opts file
# (in test file test/config-file.bats, line 33)
#   `expectedOutput | assert_output' failed
# /cygdrive/c/sbt-extras/test/test_helper.bash: line 131: warning: add_process: pid  2080 (stdin_or_args "$@") marked as still alive
# 
# expected:
# -Xmx1g
# -Dsome.prop="pass a # in a string"
# -Xss4m
# -Xms1g
# -jar
# $ROOT/.sbt/launchers/0.13.12/sbt-launch.jar
# about
# 
# actual:
# -Xmx1g
# -Dsome.prop="pass a # in a string"
# -Xss4m
# -Xms1g
# -jar
# $ROOT/.sbt/launchers/0.13.12/sbt-launch.jar
# about
# 

[pawelpanasewicz]

It looks like we have sbt-extras working under both linux and cygwin.
Travis and Appveyor managed to test it successfully.

https://ci.appveyor.com/project/pawelpanasewicz/sbt-extras/build/1.0.38
https://travis-ci.org/jawp/sbt-extras/builds/165897944

[paulp]

@pawelpanasewicz Great work! It restores a little of my faith in humanity to see a contribution of this kind. Probably I have to flip a switch somewhere to enable appveyor tests?

[paulp]

Okay, I think I have appveyor set up, but the easiest way to test that is to merge this so that's what I'm doing.

[dwijnand]

Nice one @pawelpanasewicz!

[pawelpanasewicz]

Thanks guys! Nice to read such words!