Closed JuanCaicedo closed 7 years ago
That's intended. It's the bundled version of Quixote, which is built by the full build (jake.sh
).
Normally you wouldn't see a diff after building a clean version of master, but in your case it's probably a result of you using a different version of one of our npm modules.
To avoid creating it, use the quick build (jake.sh quick
).
jake.sh quick
seems to run the linter but no tests, is that right?
As for the different versions, what would you think of using yarn for installing dependencies? It's an extra step, but we're using it for a few things at work and have had a good experience. It locks down dependency version so that the build is always reproducible
jake.sh quick
runs tests, but not if the files haven't changed. It also only lints files that have changed. (To force it to re-run everything, run jake.sh clean
then jake.sh quick
. Or just run ./jake.sh
.)
Regarding yarn, I usually lock down dependency versions by checking in node_modules
. I didn't in this case because Quixote's a published module. I'm not opposed to using yarn, but so far it hasn't been enough of an issue to warrant changing workflow.
We also check in node_modules at work, but we've found it to be a pain point as time goes on.
I'll move the yarn discussion into another issue (#48), then if there are enough drivers for it, we can move to it then
On a clean version of master, if I start the server with
$ ./jake.sh karma
and then run$ ./jake.sh loose=true
in another tab, I get a diff in the following filesThe diff in each seems to be something along the lines of this being deleted
(The diff is much bigger, but I thought it might not be worth adding all of it)
Is this intended? Do the tests perhaps build the library to test?