🇺🇸United States @capellic

Super curious! Thanks @cosmicdreams for posting this. I looked to see if there might be a thread in Slack, I looked in the #performance channel, didn't see one, so I started one. https://drupal.slack.com/archives/C4M1EV8G5/p1738413929098469

Nothing that @larowlan has created a ticket for getting it into Drupal Core https://www.drupal.org/project/drupal/issues/3453319 ✨ Add an API to support speculation rules Active

@fathershawn, I thought of this conversation when watching the Theo video that was released yesterday, "The Future of HTMX," wherein he reads through the article with same title by Carson Gross and Alex Petros found here: https://htmx.org/essays/future/. I'm thinking specifically this part of my point #2 above, "Reduce the markup payload."

Here's the section from the article below and here's a link to the time in the video when Theo introduces and discusses it: https://youtu.be/0rtpsJSLb44?si=uYkxxi5IxgxH96Au&t=2684

One part of the hypermedia ecosystem that htmx has already helped improve is template engines. When we first wrote about how “template fragments” make defining partial page replacements much simpler, they were a relatively rare feature in template engines.

It seems to me that the pursuit of lean fragments is worthwhile to realizing the full potential of the HTMX module. Curious for our thoughts!

Coming back to this ticket to note that I think it maybe a dupe of https://www.drupal.org/project/crop/issues/3200311 → . We're going to give the patch on #6 a try.

capellic → created an issue.

Thank you Marcus! And thank you and James for all the work you're doing to enable site builders with AI!

Hello, I ran into this problem this morning.

I tried applying the patch [https://git.drupalcode.org/project/ai_automator_extractor/-/merge_reques..., but I'm getting errors:

  - Applying patches for drupal/ai_automator_extractor
    patches/ai_automator_extractor-3489955.patch (Typo in dependencies, https://www.drupal.org/project/ai_automator_extractor/issues/3489955)
patch '-p1' --no-backup-if-mismatch -d 'web/modules/composer/ai_automator_extractor' < '/var/www/html/patches/ai_automator_extractor-3489955.patch'
Executing command (CWD): patch '-p1' --no-backup-if-mismatch -d 'web/modules/composer/ai_automator_extractor' < '/var/www/html/patches/ai_automator_extractor-3489955.patch'
patching file ai_automator_extractor.info.yml
patch unexpectedly ends in middle of line

patch: **** malformed patch at line 9:  

Would be great if the Dev version could be updated to correct the typo. I watched the demo of this from 10 months ago and I'm very interested to try it out! :-) Thanks for all your work on this!

To find a way to supress the admin toolbar when rendering a \Drupal\Core\Render\Plugin\DisplayVariant\SimplePageVariant

For sure! :-)

To be able to render part of a twig template, what HTMX calls a fragment.

Interesting. I can't say I'm understanding the implementation details here but the outcome is a very lean response and I like that. The example deals with buttons within a component, which is more granular than my use case, I'm basically interested in slimming down the HTML response for entity rendering (Blocks, Paragraphs).

As to your paragraph question and the general question of rendering dynamic elements that replace themselves, yes one needs an alternate source for the replacement. That's one reason for how I implemented HTMX blocks as I did. Plus inserting into or taking over the block building process seemed like a challenge. With your paragraphs, I would suggest adding an HTMX view mode that has your lazy loading attributes on an otherwise empty div, and call for the render of the paragaph in default as your HTMX request.

Good to know that my inclination of why HTMX Blocks exist in the site builder workflow as correct. :-) I'm going to think on this. I do think that
using View Modes is the key here, and my interest is making this transparent to the editor. I imagine that a site builder can designate a Paragraph to load lazily (and possible other ways using the more of the HTMX API than I'm currently thinking about). Let's assume that's the Default View Mode. Then, when the Default View Mode is rendering, I'm thinking that if it is designated to load lazily, a "HTMX" View Mode is swapped in to deliver the HTML that will then reference the View Mode that was requested. I'm thinking we can send an argument on that second request to tell Drupal to actually return the the component markup instead of the HTMX placeholder. 😬

Here are three suggestions that are top of mind for me:

1. Simplify site builder workflow. Currently a site builder must first create an HTMX block to then reference a standard block. Can this step be eliminated and instead add a HTMX configuration tab to standard blocks?
2. Reduce the markup payload. This was discussed in Slack here. You indicated that what's being returned is standard and expected. You know more about this stuff than I do but it still feels odd/unnecessary as we're hell bent in only delivering to the browser what is absolutely needed. Noting that the current state also causes the admin toolbar to appear because it is not rendered as a block. You indicated that building a new page variant would be the way to solve this but is also a big effort.
3. Paragraphs support. Many websites are built with Paragraphs. Would be great to be able to configure Paragraph to use HTMX.

Not sure how this next bit would fit into a roadmapping feature, so here it goes:

Our use case is narrow, we want to lazy load the entity, and I've experimented by writing Twig code to do the HTMX magic but I'm running into an issue where the very View Mode (Default) that is only rendering an HTMX placeholder is then needing to call the Default View Mode to actually rendering the markup... so its a self referencing loop. I was thinking that I might have a "swappable" View Mode that gets dynamically created to allow this passthrough. I imagine that this is why the blocks site building workflow requires an HTMX block which is what is placed in a region and which then calls the standard block?

Got it working with a static ID, this code ins the paragraph--type.html.twig file:

{% set htmx = create_htmx() %}
{% set path = 'htmx/paragraph/451/default' %}
<div {{ htmx.get(drupal_url(path)).trigger('revealed') }}></div>

I dig some additional banging around and I have some progress to report:

I learned that I could test for the existence of HTMX on the page by entering `htmx` into the Chrome console. If the library is loaded on the page you get the object as a response, otherwise you get an error.

I had been under the impression that `{{ attach_library('htmx/drupal') }}` wasn't working, but I just don't think I was clicking the cache clear button hard enough. ;-)

I discovered I could enter XHTML right into the Twig file, this one was fun:

<div>
    <div id="response-div"></div>
    <button hx-get="htmx/node/105/card" hx-target="#response-div" hx-swap="beforeend">
        Register!
    </button>
</div>

Once I got confident the HTMX library was loading I experimented with lazy loading the card i the screenshot above:

{% set htmx = create_htmx() %}
<div {{ htmx.get(drupal_url('htmx/node/105/card')).trigger('revealed') }}></div>

I chose "revealed" after experimenting with "load", however "load" was firing many times on the page.

I'm going to close this ticket. I hope someone else finds it useful.

capellic → created an issue.

Thanks @fathershawn,

I plan on reporting back when I've given this a try. I have time blocked next week for a little R&D. :-)

Also experiencing this problem.

Thanks @drdam, I appreciate the assistance.

that's what I would recommend. The "derivative cache" are supported by the CDN, so no need to "force drupal" to handle it.

I just tested this and it doesn't appear to be working for me as you describe.

1. Created a new node with a new image media entity.
   1. Uploaded a panoramic image so its easy to recognize when changing the focal point for the 4:3 use case.
   2. Defined the focal point on the far left of the panoramic image
2. Saved the node, viewed the node, looks fine.
3. Update the media entity to define a cropping focal point to the far right.
4. Refreshed the page/node that the media entity appears on, no change to the cropping of the image, it appears that the stored in Fastly CDN has not been purged to then grab the updates version.

Thanks @drdam,

Did you use sort of CDN or S3-like storage services ? In this case, the disabling of the flush_derivative_images, in settings are the way to fix the issue.

We are hosting on Pantheon which sits behind the Fastly CDN. Sounds like disabling flush_derivative_images is recommended with this setup?

In crop entity perspective, there is way to detect if the newly created crop is created from a "new uploaded image" or because someone change the ImageStyle configuration to adding crop in it (and a derivative image already exist for the source and must be flushed).

I've read this a couple of times and I think that you are missing the word "no" in the above sentence, that "... there is no way to detect...". Can you clarify?

As a note, I think it's necessary to question the need to have 250 different image styles in the project, just for the issue of storage space consumption (green-it / eco-design / cost / greenwashing...).

It's actually half that but point well taken. I dream of the day we can set one Image Style and Fastly's IO product will use field data to determine the optimal responsive image set.

capellic → created an issue.

capellic → created an issue.

capellic → created an issue.

Oh that's great!

I can't apply those steps right now but will open a new ticket if I have issues? Or keep this one open?

capellic → created an issue.

@danrod

I don't have anything additional to add but am also not sure what information would be useful. The spam submissions have been preserved (Webform).

Great to see this functionality returned! Thanks!!

Just got word from a client that they got hundreds of spam messages over the last 24 hours. All from `45.144.227.*`, all from Taiwan.

Following up on tagging a new release. :please:

Thank you @DamienMcKenna for this patch and all that you do! The patch works for me!

While I think the need for this patch is outdated, I just wanted to note that it caused a bug for us documented over here: https://www.drupal.org/project/paragraphs_ee/issues/3428604 🐛 Add-in-between does not work / all elements are added at the end Closed: works as designed

We figured it out. There was a patch to Paragraphs that was causing the problem.

https://www.drupal.org/files/issues/2022-06-07/SplitText-add-more-button... → on this issue https://www.drupal.org/project/paragraphs_features/issues/3282391#commen... 🐛 Text Split feature is not working with latest Paragraphs module Needs work

@stBorchert, thanks for looking into it.

Same issue here.

Thanks @YevKo, this helped me:

I found out that .. disabling .. Aggregate CSS files and Aggregate JavaScript files what leads to the site getting up again.

I was looking for a solution to this and found this ticket. Time for an upgrade! Thanks all!!

Thanks!

capellic → created an issue.

capellic → created an issue.

This thread is fairly old and now you can add attributes via *.libraries.yml

https://www.drupal.org/docs/develop/theming-drupal/adding-assets-css-js-... →

Thanks!

There were some older blog posts I've now included in the feed, the count now stands at 3.

We'll have another article coming up in early January, I'll update the ticket then.

capellic → created an issue.

Thanks @bkosborne, this is helpful.

@catch

is cloudfare caching anonymous HTML requests?

We have a cache rule setup to bypass Cloudflare cache.

Hello all,

We did our first Drupal 10 upgrade deploy last week (October 4) and noticed this error appearing. We have Sentry hooked up to this site so its pretty easy to see the frequency, user agent meta data and the stack trace. I see the user agent and stack tracking has been mentioned above, so I'll provide those details below.

We do have Cloudflare out in front of this site.

Number of errors per day

October 4: 13
October 5: 33
October 6: 44
October 7: 28
October 8: 19
October 9: 18
October 10: 27
October 11: 15
October 12: 14

Browser

1. Chrome Mobile 117.0.0: 49%
2. Google bot 2.1: 13%
3. Chrome Mobile 115.0.0: 5%
4. IE 10.0: 4%
5. Other: 26%

Interesting that Chrome desktop or Safari doesn't appear at the top of this list. This site isn't "mobile-forward" so it does stand out to me that chrome mobile is representing 62% of the browser generating the errors.

Browser.name

1. Chrome Mobile: 64%
2. Gooblebot: 13%
3. Google Chrome: 4%
4. Other: 12%

Furthering my observation in "Browser" section above.

Client OS Name

1. Android: 84%
2. Windows: 8%%
3. macOS: 2%
4. Other: 2%

Device Family

1. K: 64%
2. Spider: 19%
3. Mac: 4%
4. iPhone: 2%
5. Other: 9%

K? Huh? I can only assume that's shorthand for Android? Or is this bot traffic?

Stack trace

/core/modules/system/src/Controller/AssetControllerBase.php in Drupal\system\Controller\AssetControllerBase::deliver at line 132
In App

    // First validate that the request is valid enough to produce an asset group
    // aggregate. The theme must be passed as a query parameter, since assets
    // always depend on the current theme.
    if (!$request->query->has('theme')) {
      throw new BadRequestHttpException('The theme must be passed as a query argument');

Tell us where your source code is
    }
    if (!$request->query->has('delta') || !is_numeric($request->query->get('delta'))) {
      throw new BadRequestHttpException('The numeric delta must be passed as a query argument');
    }
    if (!$request->query->has('language')) {

Called from: [internal] in call_user_func_array

/core/lib/Drupal/Core/EventSubscriber/EarlyRenderingControllerWrapperSubscriber.php in Drupal\Core\EventSubscriber\EarlyRenderingControllerWrapperSubscriber::Drupal\Core\EventSubscriber\{closure} at line 123
In App

/core/lib/Drupal/Core/Render/Renderer.php in Drupal\Core\Render\Renderer::executeInRenderContext at line 592
In App

/core/lib/Drupal/Core/EventSubscriber/EarlyRenderingControllerWrapperSubscriber.php in Drupal\Core\EventSubscriber\EarlyRenderingControllerWrapperSubscriber::wrapControllerExecutionInRenderContext at line 124
In App

/core/lib/Drupal/Core/EventSubscriber/EarlyRenderingControllerWrapperSubscriber.php in Drupal\Core\EventSubscriber\EarlyRenderingControllerWrapperSubscriber::Drupal\Core\EventSubscriber\{closure} at line 97
In App

Called from: /code/vendor/symfony/http-kernel/HttpKernel.php in Symfony\Component\HttpKernel\HttpKernel::handleRaw

Show 1 more frame

/core/lib/Drupal/Core/StackMiddleware/Session.php in Drupal\Core\StackMiddleware\Session::handle at line 58
In App

/core/lib/Drupal/Core/StackMiddleware/KernelPreHandle.php in Drupal\Core\StackMiddleware\KernelPreHandle::handle at line 48
In App

/core/modules/page_cache/src/StackMiddleware/PageCache.php in Drupal\page_cache\StackMiddleware\PageCache::fetch at line 191
In App

/core/modules/page_cache/src/StackMiddleware/PageCache.php in Drupal\page_cache\StackMiddleware\PageCache::lookup at line 128
In App

/core/modules/page_cache/src/StackMiddleware/PageCache.php in Drupal\page_cache\StackMiddleware\PageCache::handle at line 82
In App

/modules/contrib/advban/src/AdvbanMiddleware.php in Drupal\advban\AdvbanMiddleware::handle at line 57
In App

/core/lib/Drupal/Core/StackMiddleware/ReverseProxyMiddleware.php in Drupal\Core\StackMiddleware\ReverseProxyMiddleware::handle at line 48
In App

/modules/contrib/cloudflare/src/CloudFlareMiddleware.php in Drupal\cloudflare\CloudFlareMiddleware::handle at line 185
In App

/core/lib/Drupal/Core/StackMiddleware/NegotiationMiddleware.php in Drupal\Core\StackMiddleware\NegotiationMiddleware::handle at line 51
In App

/core/lib/Drupal/Core/StackMiddleware/StackedHttpKernel.php in Drupal\Core\StackMiddleware\StackedHttpKernel::handle at line 51
In App

/core/lib/Drupal/Core/DrupalKernel.php in Drupal\Core\DrupalKernel::handle at line 704
In App

/index.php at line 19

capellic → created an issue.

Thanks for the update @stBorchert.

To be clear, I don't care what the background color is - white or black I'm indifferent. But the issue here is that the text is illegible because it is light gray on white.

This is a good point.

However, and not an argument to keep this module, but I would like a way to set the webp conversion within an Image Optimize API pipeline and not within each individual Image Style.

@DewinImagro

No, no luck. And we tried the patch on https://www.drupal.org/project/facets/issues/3282290 🐛 Combined facet active item incorrect Active without any luck either.

This might be helpful.

https://www.drupal.org/project/gin/issues/3345171 ✨ Extend the Gin offcanvas library override Active

Second embed was wrong.

Embedded images weren't working, so uploaded them.

capellic → created an issue.

For anybody else who has this problem, check your View to be sure that you aren't unwittingly excluding the field.

@gordon, love your approach here. We too want to follow your workflow of using both CKEditor 4 and 5 for a time. We'd like to be transitioned over before we upgrade to Drupal 10.

I'm having the same problem.

The way we build our sites, we have a "Common Description" field that we use for various summary view modes (Teaser, Card, etc). We also use the content from this field for og:description and twitter:description and the description meta tag. Naturally I want this to be the field that is used.

That's odd is that the widget configuration appears to support this need:

@gaurav_manerkar any luck working around this?

I'm wondering if this is related to https://www.drupal.org/project/facets/issues/3282290 🐛 Combined facet active item incorrect Active

We need to do the same thing.

In our situation we have two these two fields we need to combine:

Primary Resource Type - Reference to Resource Types taxonomy vocabulary, single-value field
Secondary Resource Types - Reference to Resource Types taxonomy vocabulary, multi-value field

A given node might look like this:

Primary Resource Type
- Type 1

Secondary Resource Types
- Type 2
- Type 3

We want to provide a single facet to the user called "Resource Type"

The node should appear if any of the terms (Type 1, Type 2, Type 3) are selected, regardless of the field.

We've been experimenting with the Combine option on one of the facets, but it doesn't appear to work.

@cianevans have you been able to resolve?

Once I remove that 'field', and added the VBO field, it ran through as expected.

Thanks for reporting this tip... helped me out today when I was in a jam.

capellic → created an issue.

I've tested this patch and it works well. Thanks!

capellic → created an issue.

added missing steps for testing

Took a closer look at the module code and I see that this integration is intended to work:

Image style class to support retina generation with Image Optimize Pipeline.
https://git.drupalcode.org/project/auto_retina/-/blob/8.x-1.x/src/Entity...

auto link formatting correction.

auto link formatting correction.

HTML markup error.

capellic → created an issue.

I'd like to see an interface option for dealing with this use case.

BTW, if anybody is looking for a way to do this via Terminus, this will work:

terminus remote:drush SITE.ENV php:eval "pantheon_clear_edge_paths(['path/to/file/filename.pdf']);"

I too have a View with an Entity Reference in the Relationships area. All was fine until I started using the Views Unpublished module which caused the View to disappear for those users who didn't have the rights to see unpublished nodes that appeared in that View. Applying the patch in #195 works for me. Thank you all for your work on this, sticking with it, and getting us here.