entering values and triggering events in tests #5

jywarren · 2016-05-26T23:33:30Z

This pull request isn't ready -- it's a placeholder.

It will incorporate event triggering to cause a horsey to appear, to lay the groundwork for horsey testing. No animals harmed, etc.

horsey appearances using new enter() test method (instead of horsey appearances using editor.value())
resolve .is(':visible') test inconsistency -- is the show/hide triggering asynchronous or something?
resolve banksy.js:84 Uncaught TypeError: Cannot read property 'before' of undefined on triggering click on menu using jQuery. Possibly switch to a crossvent event.
horsey appearances in different woofmark modes
documentation of where tests live and how to run them

jywarren · 2016-05-31T20:50:56Z

Progress, but not 100%; will continue working tonight but have to catch a plane now.

bevacqua · 2016-05-31T21:51:25Z

Have a safe 🛩

jywarren · 2016-06-01T03:48:31Z

Thanks -- got a lot of this working, and I am now near the point of not being sure if I'm failing to write a test or the test is failing. Especially on the first line each of onShow and onHideAgain, I believe these things should be true, and when I manually test them they are true. So I may just be on the tail end of debugging my event listening testing code. Do these tests look overall OK to you?

bevacqua · 2016-06-01T04:28:11Z

They look fine at a glance. Don't really have time to go deep into them now, but integration tests are a fragile thing. You'll probably need a bit of console.log-ing and a bit of luck to fix whatever subtle issues you may be running into

text entry using runCommand autocompletion click, event listeners, but callback hell refactored callback hell basic usage tested

jywarren · 2016-06-02T19:44:23Z

I just got this working, and think it's close to ready for merge, or perhaps ready if you agree. I'm going to move the last two items into a separate pull request.

There's just one oddity left -- I had to run a setTimeout of zero on line 100 to get the check for they horsey appearance to work. I'm not clear on why this is, but speculate that crossvent.fabricate may do something to pull the event out of the thread, perhaps, so my setTimeout is queuing the next test after the event fires?

jywarren · 2016-06-02T20:25:17Z

Changed my mind -- testing multiple modes made me refactor some of this code, and it made enough sense to submit it together. I also have a question -- the last test is failing because a trailing space is inserted in markdown mode that I didn't expect -- but the space doesn't seem to be inserted in wysiwyg mode. Which is the intended behavior?

Expected 'hello markdown @hodor ' to equal 'hello markdown @hodor'.

jywarren · 2016-06-08T14:11:46Z

I think the extra space may be this   on line 88: https://github.com/bevacqua/banksy/blob/master/banksy.js#L88

I see how inserting an extra space makes sense, but it's odd that it's not reflected in wysiwyg mode. Maybe the contentEditable doesn't return it somehow, or maybe wysiwyg mode runs an extra trim operation?

jywarren · 2016-07-19T14:21:59Z

spec/javascripts/banksy_spec.js

+      expect(editor.value()).toEqual('hello wysiwyg @hodor');
+
+      editor.runCommand(function(chunks, mode) {
+console.log(chunks,"expect insertion point to be at the end");


@bevacqua - I'm circling back after a few weeks of other work, and I've tracked things to here. Using console.log I can see that chunks.before and after are correct within banksy's custom appendHTML() function. But by the time I test them here (triggered by the crossvent.fabricate on line 111), the insertion point has moved to the beginning of the text, and chunks.before and after have changed. Any idea on when/where/how this might be happening? Maybe some kind of focus/refocus event in the midst of user interaction? Thanks!

I've followed chunks all the way to writeSelectionEditable in InputState and getSurface, and they keep the input text in chunks.before late in the process as I can follow them, it seems. Only when getChunks() is called can I console.log out a chunks state where the inputted text has been moved to chunks.after.

I can also visually confirm that the insertion point is moved to the beginning by interacting manually with the wysiwyg mode in the open test window, too.

Progress... I found that InputHistory.setCommandMode here is run, triggering InputHistory.refreshState, here, where I believe the InputState's start and end properties are overwritten. This happens exactly where the chunks.before and chunks.after become incorrect, and this relates to history. I want to find what's triggering setCommandMode, or why refreshState is overwriting the cursor position.

I also see that here, setCommandMode (which runs a little before the test fails) can trigger refreshState, but does so within a setTimeout command, so asynchronously. Since the test (as currently written) also uses setTimeout (to get the events to sequence correctly), I wonder if this setCommandMode call is actually getting asynchronously executed in the middle of the test.

I'm going to look at crossvent and see if we can remove the setTimeout usage from the test code.

Hmm, removing the setTimeout seems actually to allow the chunks object to retain the correct values, but causes the click event not to trigger properly, which is why I'd added it in the first place. @bevacqua - I'm really sorry to bother you, but can you offer any insight into how I might get the click event to trigger in this test without using a setTimeout(function() {}, 0)? I think if I can get that to work, I can clean up and complete this PR.

Also, I think the before/after swap issue I'm chasing may not be an artifact of the timing of the tests themselves. I'm seeing the behavior also when manually entering hi @h and autocompleting, which you can try here: https://publiclab.org/editor/

I'm looking into this again, @bevacqua -- I'm thinking that I need to perhaps set aside the integration tests for now and examine some of the lower, unit-level assumptions, maybe in horsey before I can crack this. Horsey's v4.x doesn't have a horse.anchor property, though, so it's changed a good bit between v3 and v4. Are you planning on maintaining banksy to support horsey v4 and forward?

I haven't really had the time to work on this one for a while. I'll collaborate with you in any way I can though

Egads! I worked through the changes in the horsey API for v3/v4, and the insertion point shift bug is no longer occurring in manual use. (!!)

I'm going to try refactoring these tests for horsey v4, and (not to jinx it), maybe these tests will pass? Whoa, sudden excitement.

I'm getting a little bogged down converting to v4, but making progress. Horsey's triggering of setMode is within a setTimeout deferred method, and I'm having trouble getting my tests to run after a mode change. But I think it's going to be more rewarding than writing the test suite against v3 of horsey, long term, and I was basically stuck, anyhow.

In manual use, too, I am seeing improper ordering of the crossvent.add listeners due to the deferment of setMode in woofmark. A bit of a thicket and I've asked on ponyfoo slack for some tips on this kind of code.

entering values and triggering events in tests

936d1a1

text entry using runCommand autocompletion click, event listeners, but callback hell refactored callback hell basic usage tested

jywarren force-pushed the jasmine-test-events branch from 932c471 to 936d1a1 Compare June 2, 2016 19:36

testing in multiple woofmark modes

5c0bd1b

test event debugging

0fd202c

jywarren reviewed Jul 19, 2016
View reviewed changes

jywarren mentioned this pull request Jul 19, 2016

Double anchor fix #6

Open

jywarren mentioned this pull request Aug 1, 2016

Nested lists are not handled properly bevacqua/domador#6

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

entering values and triggering events in tests #5

entering values and triggering events in tests #5

jywarren commented May 26, 2016 •

edited

Loading

jywarren commented May 31, 2016

bevacqua commented May 31, 2016

jywarren commented Jun 1, 2016

bevacqua commented Jun 1, 2016

jywarren commented Jun 2, 2016

jywarren commented Jun 2, 2016 •

edited

Loading

jywarren commented Jun 8, 2016

jywarren Jul 19, 2016

jywarren Jul 19, 2016

jywarren Jul 29, 2016

jywarren Jul 29, 2016

jywarren Jul 29, 2016

jywarren Jul 29, 2016

jywarren Aug 15, 2016

bevacqua Aug 16, 2016

jywarren Aug 16, 2016

jywarren Aug 16, 2016

entering values and triggering events in tests #5

Are you sure you want to change the base?

entering values and triggering events in tests #5

Conversation

jywarren commented May 26, 2016 • edited Loading

jywarren commented May 31, 2016

bevacqua commented May 31, 2016

jywarren commented Jun 1, 2016

bevacqua commented Jun 1, 2016

jywarren commented Jun 2, 2016

jywarren commented Jun 2, 2016 • edited Loading

jywarren commented Jun 8, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jywarren commented May 26, 2016 •

edited

Loading

jywarren commented Jun 2, 2016 •

edited

Loading