Responded to @janboddez:

@janboddez Is this going to impact your WP plugins?

Responded to @kasparsd:

Here are the customizations I applied to get it working like I want: https://kaspars.net/blog/seriously-simple-podcasting-setup

Surprised how quickly I could setup a completely new podcast on all the major distribution platforms (Spotify, Apple and YouTube) using WP with the Seriously Simple Podcasting plugin.

I recently had to use Meta apps and websites for managing ads on Facebook and Instagram. I can’t believe how bad the web and app experience is — terrible layout shifts, custom input controls that don’t support basic keyboard navigation, constant loading screens.

Responded to @vattuvarg:

@vattuvarg on the Mikrotik side you need to setup this https://help.mikrotik.com/docs/spaces/ROS/pages/26476608/Failover+WAN+Backup along with an LTE modem on a port. Consider the updated models like https://mikrotik.com/product/sxt_lte6_2023 with two SIM slots.

Responded to @oivaeskola:

@oivaeskola You’re absolutely correct! Missed it when refactoring the first example.

It’s always fascinating to see how different developers tackle the same coding problem. WordPress often prefers to "bail early" with early returns, but this can lead to inverse conditionals that I personally find harder to read. Which approach do you prefer? (see attached)

Here is the context: https://developer.wordpress.org/news/snippets/how-to-disable-heading-levels-in-the-editor/

I'm cross-posting content to all major social media platforms (via Buffer). Let's connect!

– Twitter: https://x.com/konstruktors
– Bluesky: https://bsky.app/profile/kaspars.net
– Threads: https://www.threads.net/@wpelevator

There is a beautiful blogging platform Bear. https://herman.bearblog.dev/building-software-to-last-forever/

Responded to @kasparsd:

Here is the issue: https://github.com/moby/moby/issues/25526

This week I learned that Docker Swarm has a major bug where it fails to forward the HTTP request IP to the containers. This leads to incorrect IP attribution in the logs and servers thinking the request came from an internal/private IP. No wonder there are so many attempts at creating something between Docker Compose and k8s.

I'm trying out the stale-while-revalidate approach with WP object cache — whenever the main cache entry expires, it attempts to serve a stale version and schedules an update on the shutdown hook after flush()-ing output. Almost like async PHP 😅

Turns out you don't need to extract a ZIP file to rename a directory inside it. I'm adding GitHub sync to the Update Pilot Server and it will handle this automatically. One nuance is that you have to move all entries that are under that directory individually.

Responded to @kasparsd:

The vendor name could also be the reverse hostname such as `com.vendor-name` like iOS and Android apps do for namespacing.

Here is the full report: https://gist.github.com/kasparsd/f5e1544f3eb14023b1f3c34c2ead35ef

Did you know that all WP ZIPs have a directory inside them matching the plugin/theme slug? They are used for the respective plugin or theme directories during install and update. With distributed update servers, I propose we prefix them with "vendor-name." (including the dot).

Here is the first draft for the Update Pilot Server landing page. All feedback is much appreciated! https://wpelevator.com/plugins/update-pilot-server

The Update Pilot client plugin uses the official WP core UpdateURI feature for plugin and theme updates from external sources. Can you think of a reason why the plugins and themes from the same vendor (as identified by the UpdateURI hostname) should have seperate fields for product update keys?

Here is an example of when unit tests could have uncovered the issue. WP REST callbacks are easy to test. https://www.wordfence.com/blog/2024/11/really-simple-security-vulnerability/

Today the Update Pilot client plugin updated itself from the Update Pilot Server over WP REST API 💪🚀

Looks like Jetpack Boost is now adding Photonized image URLs to post content. I believe it was never the intention and it started happening with the block editor REST calls not being is_admin().