Bad substitution when setting up new config for tps environment

[whimboo]

zip_safe flag not set; analyzing archive contents...
httplib2.init: module references file
/mozilla/code/coversheet/downloadedtests/tps/INSTALL.sh: 75: /mozilla/code/coversheet/downloadedtests/tps/INSTALL.sh: Bad substitution

http://mxr.mozilla.org/mozilla-central/source/testing/tps/INSTALL.sh#75

NEWCONFIG=${CONFIG:0:${#CONFIG}-3}

[whimboo]

The command I was using is coversheet --configfile config/config.json --pulsefile tests/pulse_messages/mozilla-central.en-US.linux64.json

[whimboo]

When I run bash downloadedtests/tps/INSTALL.sh env it also fails:

downloadedtests/tps/INSTALL.sh: line 75: ${#CONFIG}-3: substring expression < 0
downloadedtests/tps/INSTALL.sh: line 77: cd: ../../services/sync/tests/tps: No such file or directory
downloadedtests/tps/INSTALL.sh: line 80: cd: ../../tps/extensions: No such file or directory
downloadedtests/tps/INSTALL.sh: line 83: : No such file or directory
sed: can't read : No such file or directory
rm: missing operand
Try 'rm --help' for more information.

[whimboo]

It looks like that the INSTALL.sh file is broken because I'm using coversheet from within an already existent venv. So we are trying to use a venv within a venv. For testing I don't really want to install all the coversheet packages globally on my machine. @jonallengriffin any advice? Not sure how the official CI machine handles that.

[jonallengriffin]

Coversheet tries to create a virtualenv and doesn't detect if you're already in one. We should probably add that ability. For the CI, I've just installed coversheet's dependencies globally.

[whimboo]

Ok, so that is most likely a problem inside of INSTALL.sh, and should be covered on bugzilla. I think that we might also want to convert this script to Python, so we can use some nifty code. I will check back once the remaining tps issues are fixed. Otherwise whoever wants to work on it, feel free.

[whimboo]

@cosmin-malutan how does this work now? I assume it is fixed given that we moved away from the INSTALL.sh script.

[cosmin-malutan]

Yes it is, I tested the command from #10 (comment) with a pulsefile pointing to the latest build whitch implicitly uses create_venv.py instead of INSTALL.sh and as expected it passed.