Starter Content Pattern Experience Follow Up Tasks

[fabiankaegy]

In #66836, the "Starter Content" Modal was replaced with Zoomed-out mode. This Issue here is meant to track some of the follow up tasks:

Loading Performance

Loading the full editor feels slower and more "janky" with this new behavior. See attached video. There are sooooo many cumulative layout shifts happening here and the time to interactive / useful content loaded is really sub optimal.

Yes part of this is coming from the fact that currently for pages the template is shown by default and loading template parts is not a great experience either. But this is even more pronounced in the zoomed out mode at the beginning because first users get to see an "error" state saying no patterns available. And then it loads for quite a while.

CleanShot.2025-01-17.at.09.33.31.mp4

Wrong initial pattern category selected by default

As it stands right now the "All" category is selected rather than the "Starter content" one.

The default placeholder for adding content is not shown in zoomed out mode

Another recent change that was applied to the Zoomed out mode is that placeholders of blocks such as the Paragraph are no longer shown in zommed out mode. This makes it so a user cannot start writing content without starting from a pattern first.

CleanShot.2025-01-17.at.09.59.50.mp4

It's hard to exist the Zoomed out mode

This is especially true for someone that isn't familiar with the software. But even for me who spends multiple hours every single day using WordPress I am really disoriented by the Zoomed Out mode experience. Getting interrupted by this mode and first having to figure out how to get out of it / how to close the inserter is really slowing down the user experience.

[t-hamano]

I would like to suggest two issues. I don't know if either of these issues existed in the previous Pattern Modal Explorer.

Starter content reappears

If I update only the title and save as draft or publish, the inserter will appear again. It might be better not to show the starter content if the post has a title:

3c433966e6058f0744fa04ca7c75a2f9.mp4

Accesibility check

We should test a new page with a screen reader to make sure it reads well when we create it. Otherwise, users who can't see the screen may not understand what's going on. cc @afercia

NVDA will read it like this.

New post (no starter content):

Add New Post ‹ gutenberg — WordPress  document   
Add New Post ‹ gutenberg — WordPress  document
unknown

New page (with starter content):

Add New Page ‹ gutenberg — WordPress  document  blank
0 category buttons displayed.  
main landmark
Block Library  region
tab control
Patterns  tab  selected  2 of 3
Patterns  tab  selected  2 of 3
17 category buttons displayed.  
All  tab  selected  1 of 17
Page List: Cannot retrieve Pages.  
Drag and drop an image, upload, or choose from your library.  

[afercia]

I'm not sure this behavior should be forced for all users when they click 'Add new page` in the first place.
To me, this seems a bold assumption on a specific flow that may not be what users want.

I was not aware of this new experience being considered and I'm a little surprised. Anyways, I would consider this flow as a separate one, with a dedicated button other than 'Add new page`.

On top of that: has the scenario where users may be using page builders been considered> What happens in that case when users click 'Add new page`?

[t-hamano]

I'm not sure this behavior should be forced for all users when they click 'Add new page` in the first place.

This starter content is enabled by default, but can be disabled in the settings modal:

[stokesman]

This starter content is enabled by default, but can be disabled in the settings modal:

I was going to mention the same thing (added in #65026). It helps but, for me, isn’t a complete resolution. It seems hard to discover and even once you learn about it, there may be occasions where the starter content flow is welcome but it’s clunky to have to open preferences and reenable it.

Anyways, I would consider this flow as a separate one, with a dedicated button other than 'Add new page`.

This was suggested #56181 (comment) and I’m very much in favor of the idea. It can’t be done as straightforwardly for the Post editor but maybe it’s possible.

[t-hamano]

In #66836, the reason for replacing the starter pattern with an inserter panel was that it was "super annoying". However, replacing it with zoom out mode and inserter seems to have a bigger side effect to me.

Perhaps it's worth considering reverting to the previous modal.

Personally, I would suggest adding a "Skip" or "Don't show again" checkbox in the modal, as suggested in this comment. That is, "skip" is a temporary action, while the "Don't show again" checkbox means to invalidate the value in the settings store.

[fabiankaegy]

As for right now I am also in favor of reverting and then adding a "don't show again" option to the modal. We can try to fix all of the outlined issues with Zoomed out mode in the ~month we have left but I think that is too ambitious for the amount of people we are working on it currently

[yogeshbhutkar]

The default placeholder for adding content is not shown in zoomed out mode

@fabiankaegy, I believe this was intentionally applied #68690. Do we revert it?

Edit:
Also, the placeholder text becomes too small when zoomed out if enabled.

See ref:

Also, we did hide the placeholder, but the slash menu is still visible and well accessible.

[t-hamano]

I noticed that if we decide to use the modal, we run into another issue.

In the site editor, when you create a new page draft, you get a snackbar that says "{Page} successfully created.":

But the starter content appears in front of that snackbar.

This issue is probably related to #68728 as well.

[carolinan]

There are sooooo many cumulative layout shifts happening here

This is so distracting!
It took me many months to get used to the pattern modal opening. I found that it slowed me down but I eventually got used to it. But this new layout shift is not only annoying but uncomfortable. I almost get dizzy.

[carolinan]

I have not worked on this part before but I can try looking into

Wrong initial pattern category selected by default
As it stands right now the "All" category is selected rather than the "Starter content" one.

[Mamaduka]

The new "inserter starter" content should also account for the viewport size. It's an odd experience when you create a page on a small viewport and only see the expanded pattern inserter.

[Mamaduka]

Started working on PR (#68852), which should resolve the following items.

• It's hard to exist in the Zoomed-out mode - closing the inserter will exit the mode. I'm not sure if we can make it "easier" for this feature.
• Starter content reappears - starter content won't reappear unless you reload the editor while the post content is empty.

Remaining items

• Loading Performance - this needs some exploration; maybe we can trigger zoomout mode simultaneously with the opening inserter.
• The default placeholder for adding content is not shown in zoomed-out mode - introduced in Remove placeholder of default paragraph when it's the only block and canvas is zoomed out #68106. We could show an insertion point for zoom-out mode here instead.
• Wrong initial pattern category selected by default - I see that there are two PRs, which will help with reviews and feedback.

[stokesman]

For me it seems a big obstacle to the best possible UX is that the presence or absence of "Starter content" can’t be determined initially. Both the modal or the new "assembler" experience are impacted by it and forced to have suboptimal UX for one of the cases (presence or absence of stater content). I’ll add more about how each experience is impacted.

The modal only appeared once starter content became available and this seems ideal in the case of no starter content. However, with starter content, this is what made its appearance sometimes feel "late" and especially disruptive—the editor can be fully loaded and interactive before then. See the video in #63947 as a prime example of this.

The new "assembler" experience engages regardless of whether or not starter content will be available. This is great as it prevents a disruptive "late" engagement. Without starter content, it would seem ideal if it didn’t engage but that can’t be done without opening up the potential late engagement. Thus the problem shifts to category selection. The rest of this assumes a fix for the initial category is in (like #68806). When the "Starter content" category isn’t available, no category will get selected and result in an empty pane where the pattern previews would be expected. In such case it would seem obvious to fallback to another category. Yet that can’t be done without possibly falling back too soon in the case "starter content" is "late" to populate (at which point it’s probably too disruptive to then change the category selection).

From my limited understanding, I think having the presence/absence of starter content be known when the editor loads would require some PHP work and I'm not even sure it’s possible. I feel strongly it would be worthwhile if so.

[afercia]

Determining the presence/absence of starter content would allow to provide two separated user flows. To me, forcing a specific flow that may or may not be what users want seems a big assumption. I'd tend to think these two flows should be separated:

• Add new page
• Add new page with starter content

[andreawetzel]

Noting that loading patterns means a lot of network requests to load images, css and js files. I've been avoiding adding theme patterns that include Block Types: core/post-content to avoid the patterns loading on new page creation for the performance reasons in addition to finding the Modal to be distracting. I agree that if the new page flow of opening the pattern pane remains, defaulting to a category where minimal assets are loaded would be better than landing on "All"

[stokesman]

I found another bug in the new experience relevant only in the Site editor.

If you have an starter-eligible entity selected (shown in the preview) and then use the edit action (pencil icon) to edit one which is not eligible, the experience engages unexpectedly.

assembler-false-positive.mp4

[carolinan]

As for right now I am also in favor of reverting and then adding a "don't show again" option to the modal

+1

[Mamaduka]

If we decide to revert, can we agree on the items that must be resolved when the feature is reintroduced? We could use this issue as a living document.

[carolinan]

I am not sure it should be reintroduced. I understand the irony but I don't think it was clearly and objectively described why the modal needed to be replaced.
But your list of remaining items from last week is still valid.

[t-hamano]

If we decide to revert, can we agree on the items that must be resolved when the feature is reintroduced? We could use this issue as a living document.

Does this mean that for starter content insertion, we will use the modal inserter as of WP 6.8, and aim for the ideal zoom out inserter in the future?

[carolinan]

(I think I am missing something about the difficulty to combine the pattern modal with the zoom out. Its a blank page and you can only insert one pattern before the modal closes, so where you place it in the zoom out doesn't matter)

[Mamaduka]

Does this mean that for starter content insertion, we will use the modal inserter as of WP 6.8, and aim for the ideal zoom out inserter in the future?

@t-hamano, if we decided to revert, then yes. That's would be the plan.

(I think I am missing something about the difficulty to combine the pattern modal with the zoom out. Its a blank page and you can only insert one pattern before the modal closes, so where you place it in the zoom out doesn't matter)

Right, Zoom out doesn't make sense for the modals.

[stokesman]

can we agree on the items that must be resolved when the feature is reintroduced

Though perhaps not required, ideally we’d arrive at an answer to your question #66836 (comment) on whether the experience should engage for unsupported themes. If it should not engage then is there no starter experience for such themes or would it fall back to the modal? If it should engage, is it okay that the mode/view is otherwise unavailable through the UI?

[carolinan]

What is an unsupported theme?

[Mamaduka]

I chatted with @ellatrix a couple of days ago; she agrees that the experience isn't ideal without planned changes to the Pattern inserter panel. Considering that and additional issues discovered after merging the initial PR, it might be better to revert to the Starter Content model (see WP 6.7).

What happens when #66836 is reverted?

• Revert resolved this issue and re-opened Assemble new pages with zoom out #63865.
• Gather and share recently discovered bugs in Assemble new pages with zoom out #63865.
• Reverent PR should also include the "don't show again" option suggested in Patterns: Provide a way for users to opt out of the 'Choose a Pattern' modal when adding a page #56181 (comment).

Please raise a virtual hand if you want to work on revert PR. Otherwise, I can get to it later this week.

[stokesman]

What is an unsupported theme?

TT1 is an example but I think some block themes too. I’m unfamiliar with the details but #67232 and the issue it closed should provide them. Saying unsupported is my word choice and maybe not the best.