Add task options groups without LLM use

[sashinexists]

This is a reimplementation of #1820 without any use of LLMs.

The implementation is completely different and leans on the HTML details element for the collapsibles, as a result it is more accessible and works better with only keyboard input. It is possible to expand or collapse groups by tabbing between them and pressing enter or space.

It has also been tested with a screen reader and and it is possible to tab between the groups and it will read the group names. This was not the case with the AI version.

Here is a list of changes I have put together:

Main

• Task Options now display in groups in line with specification
  • each group is a HTML
    Details

    element which will expand and collapse on click
  • appropriate font-awesome icons have been used for each group
  • There is an arrow icon at the leftmost side of each group header that will animate when opening a group

Search

• When searching any groups or subgroups with 0 results with automatically be hidden
• When the search is toggled open the background of the search button changes to indicate it is open
• Searchbar text placeholder has been added "Filter options by Search"
• Button title "Filter options by Search" added, viewable on mouseover
  • accessible by screen readers

Legacy Flat View Toggle

• Toggle between current grouped view and legacy flat view
• Toggle button changes colour when in legacy flat view
• Button title is "View options in legacy flat view (no groups)" viewable on mouseover
  • accessible by screen readers

Expand/Collapse button

• If no groups are open, this button will expand them all
• If any groups are open, it will collapse them all
• the icon will change depending on whether any groups are open or not
  • the button title which displays on mouse over will also change between "expand all groups" and "collapse all groups"
• If legacy flat view is toggled, this button will disappear

[pierotofy]

Thanks @sashinexists, but this looks like a modification on top of the current master branch, which already includes AI changes. The modifications are based on top of what is already AI-assisted code. It's not a clean room implementation so defeats the purpose of this exercise. The code also looks suspiciously AI generated. How can you claim "The implementation is completely different" ?

Currently this also breaks theme colors.

@smathermather this is not going to work and is an exercise in futility. I urge you to once again try the tools and look at what other mature projects are doing. For example review sensible policies like the one at https://github.com/cloudnative-pg/governance/blob/main/AI_POLICY.md

[smathermather]

By way of background, I provided screen shots and a hosted version of the latest changes inclusive of AI so Sashin could visualize the changes. I also provided a list of the hierarchy / grouping of options, with some changes to order for first priority options (mostly it was quite good in the existing pull request).

As per our previous conversations, the intent was to look at the product and replicate and improve, where possible, a la a pc clone or API clone.

Thanks @sashinexists, but this looks like a modification on top of the current master branch, which already includes AI changes.

Per what we discussed, Sashin, this does need to be based on the pre-1820 commit, or rather exclude the 10cac52 commit if there are commits after 1820's merge.

Currently this also breaks theme colors.

Sorry @sashinexists, I missed that in review. Please address.

I look forward to a (AI free) PR.

Glad we were able to put together one. Happy to continue the conversation around responsible AI use elsewhere. I have seen similar ones across OSGeo and HOT related repos. Useful to follow this conversation, if you aren't already, as it's not just a pull request with language, but a full list of challenges and context, as well as the start of an open-eyed discussion of them. Let's finish this experiment and continue the conversation.

---

As far as the pull request in question, it looks like it has some advantages over the existing AI written one with respect to accessibility as far as screen reader and keyboard support, in addition to the original intent. And I really appreciate that not only does this honor the thoughtful work done by @thadwald, but extends and improves upon it by virtue of taking a fresh look.

[pierotofy]

A clean room implementation is called that way because the person in the "clean" room never gets access to the original code (only receives specifications from the other room).

Any new change/edit, even a rewrite at this point, is a moot point; the code has already been seen and will influence future code changes.

[spwoodcock]

Thanks @sashinexists, but this looks like a modification on top of the current master branch, which already includes AI changes. The modifications are based on top of what is already AI-assisted code. It's not a clean room implementation so defeats the purpose of this exercise. The code also looks suspiciously AI generated. How can you claim "The implementation is completely different" ?

Currently this also breaks theme colors.

@smathermather this is not going to work and is an exercise in futility. I urge you to once again try the tools and look at what other mature projects are doing. For example review sensible policies like the one at https://github.com/cloudnative-pg/governance/blob/main/AI_POLICY.md

I really like the CNPG usage policy - thanks for linking!

By way of background, I provided screen shots and a hosted version of the latest changes inclusive of AI so Sashin could visualize the changes. I also provided a list of the hierarchy / grouping of options, with some changes to order for first priority options (mostly it was quite good in the existing pull request).

As per our previous conversations, the intent was to look at the product and replicate and improve, where possible, a la a pc clone or API clone.

Thanks @sashinexists, but this looks like a modification on top of the current master branch, which already includes AI changes.

Per what we discussed, Sashin, this does need to be based on the pre-1820 commit, or rather exclude the 10cac52 commit if there are commits after 1820's merge.

Currently this also breaks theme colors.

Sorry @sashinexists, I missed that in review. Please address.

I look forward to a (AI free) PR.

Glad we were able to put together one. Happy to continue the conversation around responsible AI use elsewhere. I have seen similar ones across OSGeo and HOT related repos. Useful to follow this conversation, if you aren't already, as it's not just a pull request with language, but a full list of challenges and context, as well as the start of an open-eyed discussion of them. Let's finish this experiment and continue the conversation.

As far as the pull request in question, it looks like it has some advantages over the existing AI written one with respect to accessibility as far as screen reader and keyboard support, in addition to the original intent. And I really appreciate that not only does this honor the thoughtful work done by @thadwald, but extends and improves upon it by virtue of taking a fresh look.

Would love to chat some time about this. There are of course many worrying implications of AI generated code for open source maintainers, but I think the tech industry as a whole overlooks some of the equally important issues such as cultural bias, IP theft, exploitative labour, and global digital inequalities. I don't want to hijack this thread by discussing all that, but would love to discuss strategies that can minimise damage on all fronts, before diving into the AI contribution shitstorm head first. I welcome input from anyone on the linked PR if of interest, to try and navigate 🙏

[smathermather]

I don't want to hijack this thread by discussing all that, but would love to discuss strategies that can minimise damage on all fronts, before diving into the AI contribution shitstorm head first. I welcome input from anyone on the linked PR if of interest, to try and navigate 🙏

I was hoping for an invitation. :) I'm really glad you started that conversation, as I'm very interested in both the promise and challenges. And beyond inevitability / sensibility arguments, which have both consent and market implications that make me uneasy, I would like to see deeper strategies to minimize damage and don't structure a techno-Taylorized environment for dev work on our tools.

Any new change/edit, even a rewrite at this point, is a moot point; the code has already been seen and will influence future code changes.

As I understand it, he was looking at the conversation but not the code here, and the above mentioned implementation and notes, but again not the code.

I think there's some confusion as the result of him trying to resolve conflicts post-hoc, and then of course analyzing what he did differently (which is kind of cool but gives the appearance of him a priori knowing the existing code).

But, let's wait for him to confirm where he started on his local repo.

[smathermather]

@sashinexists reports back he split off at the following commit which was prior to the merge of 1820:

commit a4501835a0544c1bcb3acca61e376e9e3a6a4b0b
Author: Piero Toffanin <pt@masseranolabs.com>
Date:
Sat Jan 31 18:59:02 2026 -0500
     Upgrade pilkit

@pierotofy -- How best to proceed? Is it best for Sashin to git revert just the commit merging 1820?