Bokehjs v07

[samuelcolvin]

migrating julia to work with Bokehjs v0.7

[coveralls]

Coverage increased (+0.02%) when pulling f23d206fb8e9f6ab58079325e4861f60adfe15bf on bokehjs-v07 into bfb7e6a7ac80a649cb3dc0c626d5efa1ef0bc9ce on master.

[coveralls]

Coverage increased (+0.22%) when pulling 6ead2347a1de04145e31647b1edc1b3161f31873 on bokehjs-v07 into bfb7e6a7ac80a649cb3dc0c626d5efa1ef0bc9ce on master.

[bryevdv]

@samuelcolvin this is great to see! FYI the "glyphspec" change is hopefully the last substantial change to the JSON protocol, so you shouldn't anticipate having to make more low level updates like this anytime soon. The only change at all I can think of is probably adding versioning to the JSON protocol but that will be small, and also we will make sure to do that in a back-compat sort of way.

If there is anything we can do to help out, let us know, Also if you want us to help promote any kind of announcement from @BokehPlots, or if you would like to put any content on the Bokeh Vine or Youtube accounts, let us know that too.

[samuelcolvin]

Sorry it's been so long, normal life rather got in the way.

Once I worked out what's going on the glyphspec change makes a lot of sense, it's cleaned up my code a lot.

I'm going to merge this as it's an improvement on current master, I need to improve datetime support and add a few better example; then I think we're really for another release of a Julia package. That might be a good point to bang the drums.

The only other question I have is about testing. Currently my tests just run the Julia which obviously doesn't prove that the plot has "worked". Do you have an easy example of selenium testing or similar I could slot into travis to check that a plot actually shows up?

[coveralls]

Coverage increased (+0.31%) when pulling 700a8ea8ecae0272150148ea7cacb4a5ab3347de on bokehjs-v07 into bfb7e6a7ac80a649cb3dc0c626d5efa1ef0bc9ce on master.

[bryevdv]

We are actually working right now on deploying a Seleniumn test framework, but I'm not sure I'd expect anything concrete to be able to show of for at least a month or so. We do currently have some capability for static image comparison that uses PhantomJS. You can see the test driver script here:

https://github.com/bokeh/bokeh/blob/master/examples/test

As well as the JS code that gets used to generate the PNG:

https://github.com/bokeh/bokeh/blob/master/examples/test.js

We do also compute an image diff and upload a report to S3, the code that does that is here:

https://github.com/bokeh/bokeh/blob/master/examples/test#L646

Note that we aren't actually failing any tests based on the diff yet, it's not quite robust enough yet. But it's still nice to see all the results in one place. Here is one of the reports, for example:

https://s3.amazonaws.com/bokeh-travis/0.7.0-76272bc/report.html

Hopefully some of that might be useful. Will definitely let you know when we have more to say about Selenium test. Alternatively, if you have Selenium experience, I am happy to to put you in touch with the person that's currently working on it if you have any advice or insights or questions.

[coveralls]

Coverage increased (+0.31%) when pulling 700a8ea8ecae0272150148ea7cacb4a5ab3347de on bokehjs-v07 into bfb7e6a7ac80a649cb3dc0c626d5efa1ef0bc9ce on master.

[samuelcolvin]

Discussion of testing moved to #9.