[WIP] Add HTML5 attributes to buttons

[stevepiercy]

DO NOT MERGE.

See #389

[stevepiercy]

I need some help with getting the Travis-CI to run successfully.

It appears that since the last time master was built successfully 11 months ago, something has changed with Travis-CI that now causes master builds to fail for py37 and py38:

The command "sh -e /etc/init.d/xvfb start" failed and exited with 127 during .

On every version of Python, this PR fails.

Once I get over that problem, then I'll write a test for this feature.

[stevepiercy]

Little more digging. From https://github.com/Pylons/deform/blob/master/.travis.yml#L46-L51

# For Selenium
# https://docs.travis-ci.com/user/gui-and-headless-browsers/
# http://stackoverflow.com/a/34703838/315168
before_script:
  - "export DISPLAY=:99.0"
  - "sh -e /etc/init.d/xvfb start"

This is way over my head.

When I look at each of the failures, it looks like Xenial is used for all platforms. Help!

[ericof]

@stevepiercy Help is coming, I will look into it in the next few days.

[digitalresistor]

@stevepiercy https://benlimmer.com/2019/01/14/travis-ci-xvfb/

[stevepiercy]

@bertjwregeer DuckDuckFu is strong in you. That got me over the first hurdle.

I'm working on fixing the failing functional test.

@ericof Looking ahead, what can I do to help cut a release for deform? Do I need to do anything in deformdemo, too?

[stevepiercy]

This PR has nothing to do with the failing test in deformdemo. When I manually simulate the failing test code by running deformdemo locally and clicking per the steps in this test method, I verified that all the necessary ids and texts show up. I don't understand why the automated test fails.

Perhaps the form submission takes longer to render the error message in error-deformField3 than the step wait_to_click allows? The test code seems to handle this situation, so I'm at a loss what else to try next.

[stevepiercy]

@ericof do you have any suggestion for how to fix the one failing functional test? I am stumped.

[ericof]

@stevepiercy I've been trying to reproduce it locally as well.

[stevepiercy]

@ericof how about merging this PR and cutting a release, since the failing test is not related to this PR? Then we can open a new issue for the failing test.

Meanwhile what do you do locally to run the test? I couldn't figure out any other method than running the deformdemo and manually clicking through it in a web browser. If I try an automated method, maybe I can spot the problem.

[ericof]

Merging and fixing the test after the release