Edit collect logs form cypress refactor

[asirvadAbrahamVarghese]

PR that refactors Cypress tests for Collect Logs edit form also include review comments changes from #9508

@miq-bot assign @jrafanie
@miq-bot add-label cypress
@miq-bot add-label test

[asirvadAbrahamVarghese]

This is also prone to sporadic server-related failures, as we’ve seen with the schedules form:

[asirvadAbrahamVarghese]

• a5142fd

comment - When #9504 gets merged, we can use the helper function here.
comment - Note, checking for "saved" in the message is probably all we need.
comment - Also, we can merge this now and open a new PR to update to the helper function cy.expectFlash('success', 'saved')

• e9b091e

comment - Note, the hardcoded 2 in these intercepts here and the tree select feel brittle to me. I wonder if we need to be this precise with the intercepts. Can we use glob pattern matching or are there similar POSTs that occur that we have to avoid intercept/waiting on?

[jrafanie]

Can this be used to click:

1. collapsed accordion (to expand it)
2. expanded accordion (to collapse it)
3. select any node in the accordion tree
4. anything else?

Depending on how we intend to use it, we might want to give it a more intention revealing name.

[asirvadAbrahamVarghese]

Can this be used to click:
collapsed accordion (to expand it)
expanded accordion (to collapse it)
select any node in the accordion tree
anything else?

So far, I haven't encountered a scenario where we need to collapse an already expanded accordion item. However, I've seen that behavior specifically with the heading accordions, which are already handled by the existing accordion command.

The intention of this command is:

• To expand a collapsed accordion item by clicking on span.fa-angle-right, if it's present.
• If span.fa-angle-right is not present (which means the item is either already expanded or doesn't have expandable content), then we simply select the list item.

Depending on how we intend to use it, we might want to give it a more intention revealing name.

Would toggleListItem be a better name? Given that it's not widely used, updating the name shouldn't be a problem.

[jrafanie]

This is specific to accordions so I think selectAccordionItem sounds good to me. Naming is hard but I took "select" from your description (expanding is a side effect you're not directly concerned with... only that it happens when you select). Accordion is generally what we call things here. I think we have lists in lots of places so accordion is a more specific name for it.

[jrafanie]

Imagine a tree in an accordion with 200 nodes.

Depending on what we're able to click in the accordion (toplevel accordion, nodes in the tree), we might want to be able to scope things to any arbitrary place in the tree so we don't have to loop through 200 nodes to get to the last one. In other words, if we can select any node in the tree, can we provide a parent id to do the scoping and not just the parent accordion id?

This is a great idea and I can see us being able to do iterations of improvements as we find ways to normalize pages and traverse them consistently in cypress.

[jrafanie]

The looping looks like it could get expensive so I'm being overly cautious.

[jrafanie]

cc @GilbertCherrie @elsamaryv What do you think?

[asirvadAbrahamVarghese]

My intention with the parentAccordId parameter was to scope the search to a specific parent container (like the one shown in the image), and then iterate through the list items within that section to either expand or select the matching item.

we might want to be able to scope things to any arbitrary place in the tree so we don't have to loop through 200 nodes

Yes, as you mentioned, if the tree is large (as shown below), having to search through the entire structure could definitely become a pain point.

can we provide a parent id to do the scoping and not just the parent accordion id?

But in that case, we don’t really have a parent container to scope the iteration. Instead, we're mapping <li>elements with data-node-id, so I might need to figure out how to scope the search based on data-node-id. Do you think that would be a reliable approach?

[asirvadAbrahamVarghese]

If we're okay using { force: true }, we could actually drop this change and stick with that approach. Once the heading accordion is expanded, all the child nodes are already in the DOM, they're just hidden.

So using { force: true } would still allow us to click the target element - cy.get('.list-group-item').contains(name).click({ force: true }); name can also be regex pattern, allowing for partial matching (e.g: cy.accordionItem(/^ManageIQ Region/);).
What do you all think?

If that's not the case, then instead of relying on data-node-id, we should consider implementing something similar to the menu command - where we pass an array representing the intended tree path(The array can include strings or regex patterns, allowing for partial matching as well)
In accordion list items case ['ManageIQ Region: Region 0 [0]', 'Roles', 'MainRole-user'], with 'MainRole-user' being the deepest child node we want to click.

[jrafanie]

Good question. I'm not sure. If there isn't a good selector we can use to find the right tree node by substring, we might need to change the code (if we can) to better select the correct node. I'm concerned with looping through all of them in the parent accordion. cc @GilbertCherrie @elsamaryv thoughts?

[jrafanie]

This is also prone to sporadic server-related failures, as we’ve seen with the schedules form:

sorry, I can't read the error in the screenshot. Is that the session issue losing the tree info? undefined method to_sym for nil or undefined method split for nil: #9515?

[asirvadAbrahamVarghese]

sorry, I can't read the error in the screenshot. Is that the session issue losing the tree info? undefined method to_sym for nil or undefined method split for nil

Yes a similar one, Data undefined method downcase' for nil [ops/accordion_select] (replaced the screenshot)

[miq-bot]

Checked commits asirvadAbrahamVarghese/manageiq-ui-classic@1d5b2a8~...a5142fd with ruby 3.1.7, rubocop 1.56.3, haml-lint 0.64.0, and yamllint
0 files checked, 0 offenses detected
Everything looks fine. 🍰