WordPress Plugin Mirror Downloader (Proof of Concept)
github.com
github.com
Based on Cloudflare R2 S3 object storage costs for 33GB of WordPress plugins, for a private WordPress plugin mirror would cost less than US$0.35/month though will rise as new plugin versions get downloaded and stored and if under generous R2 Class A/B write and read operation free monthly quota, the R2 operation costs would be free. The Cloudflare Worker costs based on past 24hrs metrics was 765K Worker requests at median of 2.1ms CPU time. So with 100K Worker requests/day free, cost would essentially be the $5/month Cloudflare Worker subscription fee as overages won't apply under their generous paid US$5/month subscription plan includes up to 10 million requests per month at average 10ms average CPU time. All up, it would be US$5.35/month starting costs which will rise as number of WordPress plugins and Worker usage increases as outlined here. Cloudflare CDN bandwidth is free of charge, so no need to worry about egress bandwidth fees.
free of charge, until sales shows up: https://news.ycombinator.com/item?id=40481808
IIRC that also goes for the other "self serve" plans, so includes paid plans. You need an enterprise plan (probably $5k+/m) to distribute non-Web-page-related files without violating the TOS. People often get away with it as long as they don't go too wild, but you're in even riskier territory than one already is using non-enterprise Cloudflare.
You can use R2 for multiple scenarios, including but not limited to:
Storage for cloud-native applications Cloud storage for web content Storage for podcast episodes Data lakes (analytics and big data) Cloud storage output for large batch processes, such as machine learning model artifacts or data sets
What if I instead buy a cheap 10TB hard drive and crawl at a more relaxed rate on a commodity PC at home? The cost goes way, way down after the initial investment, to near zero.. after one year the amortized H/W costs would be around $30 per month.
Additionally, 1 gigabit+ FTTH unmetered symmetric fiber Internet lines are ever more available at a competitive price.
8 gigabit unmetered symmetric fiber for 50e a month
500e initial one time purchase for a ryzen 6800u in mini desktop factor form with 32 gb of memory and 512 nvme gb drive.
Proxmox installation.
Electricity at 0,017 kwh with a 15w cpu.
You can now hosts dozens of such projects. Upgrades for later include : an UPS for eventual power cut or surge, a small NAS for mass storage, a second mini desktop for Proxmox redondancy and replication.
Since fiber lines have becomes the norm self hosting stuff at home has been a breeze and a joy for small projects.
Extra bonus : jellyfin, pihole, home assistant, and all sort of other goodies can easily get a spot on that proxmox.
There’s also a plugin that connects to WordPress.org by default but falls back to Morpheus if needed.
We’re still working on improvements. Right now, we only support English-language updates, but we’re adding more languages and features. Our goal is to support the entire API.
Bluehost operated a plugin mirror for a while, and published all their code.
Though still limiting as you can only checksum files after plugin installation. You probably need a staging WordPress install to verify mirror plugins before rolling out for production usage
Apparently [0] John Gilmore [1] said it.
It is far easier and more likely that the community around WordPress would rally around replacing WP.org than starting over from scratch with a new toolkit. Some people might flee WP altogether, but the majority of people and companies just want to get things done - not seeing them having a lot of tolerance for or interest in having to start again from square one.
To be fair, the first and best solution most people in these parts have is "Fork it, I'm going to make my own Linux distro." or something along those lines.
No, they don't get very far but I will credit their consistency.
Also, Wordpress doesn't scale well (without getting expensive), the code is bad, it is poorly managed, security issues are all over the place. It's about time we got some alternatives.
There is no reasoning with people like this.
I guess you MVP would be something that could do simple brochure websites and be easy to extend. Design it so it is easy to install for individuals or web hosting companies. Maybe plan ahead for things like multi-site, blogging, eCommerce. etc
With all the competition and even the MVP is going to be a lot of work, especially since the average user won't be very technical.
Has anyone tried to use Wordpress as a static site generator, with a few dynamic-content functions moved to Javascript against an API for alternative open-source server (or even another Wordpress instance)?
So this is the year 2024, and something amazing called a CDN has been invented. PHP does not get executed on every pageview, nor does a Mariadb connection get burned. If you are building a website or web application at scale with any framework or platform, a big part of your job is to architect it such that everything which can be cached, is. You may also employ tools like Varnish or transients. The underlying design principles are universal and whether you're building on top of WordPress or something else doesn't matter that much - in fact many experienced WordPress developers are accustomed to serving multiple millions or tens of millions of users per month, so the ways for making this work are pretty mature.
I’ve worked extensively with Wordpress; that’s precisely why I hate it so much. It needs to die.
The proper solution, 20 years from now, will not descend from this branch.
Mind you when I made my original statement I had in mind websites that handle millions of monthly uniques and are somewhat dynamic in terms of the content they publish - for smaller scales WP needs basically nothing else and just works, but this scenario is really quite easy with WP if you employ common caching strategies, and is probably impossible with the typical "I wrote this script that generates static HTML last weekend and therefore it's a revolution in CMSes" solution
Unlike PostgreSQL, MySQL was specifically designed to handle a very large number of short-lived connections. A properly configured MySQL daemon can handle tens of thousands of connections per second. If you're still opening and closing too many connections for that, MySQL supports using Unix domain sockets instead of TCP, and PHP supports persistent connections -- a sort of ad hoc connection pooling system.
There are many issues with WordPress and its ecosystem, for sure, but database connection handling isn't one.