Closed MarkRivers closed 4 years ago
I rearranged the order of things a bit and tried to match it as far as possible to the Travis-CI configuration setup.
The phases install
, build_script
, test_script
and on_finish
are executed in this order. You're free to distribute everything that needs to be done onto those phases.
The new layout of the example (today in my fork, some time later in the main repo) is installing additional stuff at the top of the file
install:
# fetch submodules (like ci-scripts)
- cmd: git submodule update --init --recursive
# for the sequencer
- cinst re2c
and puts the prepare
, build
and test
below the matrix definition:
build_script:
- cmd: python .ci/appveyor/do.py prepare
- cmd: python .ci/appveyor/do.py build
test_script:
- cmd: python .ci/appveyor/do.py test
But you could also just put the installation of re2c
above the prepare
step that builds the prerequisites - including the sequencer if configured that needs re2c
.
Not defining a version for BASE
at all (neither in the .appveyor.yml
configuration nor through a setup file) will have the scripts fall back to using a branch called master
, which in the case of the Base repository does not exist.
Following @ralphlange suggestions above I was able to fix the asyn .appveyor.yml file to build OK.
It seems to be succeeding on the unit tests with static builds but failing 4 out of 54 tests with dynamic builds. Need to track that down.
There seem to be several problems with the .appveyor.yml.example-full example.
It contains two install: sections, lines 19 and 116. appveyor throws an error when it finds the second one.
I commented out the first one and moved this line into the second install: section:
That allowed base to build. However, when building sncseq 2.2.8 I got this error:
It seems like re2c has not been installed. Is this a problem with the order of things?
In a previous build I got this error when building mingw
Is this because neither SET nor BASE is defined for the mingw builds in the .appveyor.yml.example-full file?