XF 2.4 XenForo 2.4 status and what's new under the hood?

Where are we?​

TL;DR: We're working hard to release XenForo 2.4 ASAP, but it's taking longer than expected due to scope changes and strategic decisions to wait for certain upstream developments that will benefit the long-term roadmap. Here's an analogy to explain why:

Software development is like planning a cross-country expedition with multiple destinations.

When you set out for version 2.4, you're not just driving to the next town over. You're charting a course through unknown territory with several strategic stops planned along the way - each representing a major milestone or feature release.

But the challenge is the landscape keeps changing along the journey.

• New roads open up (better technologies emerge)
• Bridges get washed out (dependencies break or become obsolete)
• You discover scenic routes that would benefit all future travellers (opportunities for architectural improvements)
• Weather conditions shift (market demands or user needs evolve)
• Your vehicle needs unexpected maintenance (technical debt must be addressed)

You can't just focus on reaching the immediate next stop. You must consider how each decision affects the entire journey ahead. Taking a shortcut to reach 2.4 faster might leave you stranded when trying to reach 3.0, 4.0 or even 5.0.

This is why scope changes occur: experienced developers are constantly recalibrating the route based on new information, ensuring the expedition can successfully reach not just the next destination, but all the strategic waypoints that follow.

The delays aren't detours, rather they're course corrections that keep the long-term journey viable.

To be slightly less cryptic, these are some of the specific challenges we have faced along the way:

A new Tiptap version is coming​

When we announced that Tiptap is coming to XenForo 2.4 it was 95% complete, and we then took a bit of a pause to work on other projects, which we have talked about since and will be discussing in this thread. Since then, Tiptap have announced Tiptap V3 which is currently in beta. Given how core the editor is to the forum experience, it makes a lot of sense to ship XenForo 2.4 with Tiptap V3 rather than Tiptap V2 as originally planned. While the changes involved are not too extensive, we also don't want to ship 2.4 with a dependency that is still in beta and subject to change. While we are not planning to wait for Tiptap V3 to be stable, necessarily, we do at least want to give it a little bit more time so we have a higher degree of confidence that we're shipping a stable editing experience.

We started talking about a rewrite (again)​

While this is not currently the direction we've decided to go in, it's responsible for us to at least consider all routes available to us to help us reach our destination.

After nearly 8 years since the release of XenForo 2.0, many of the technologies we use are showing their age, many of the decisions we made have started to slow us down more than we would like, and as a framework, XenForo becomes a less productive framework to work with. The solution to this problem can be to start from scratch, but we have ultimately decided that this is not something we need to do at this stage.

Instead, over the next few versions, including 2.4, we will be attempting to make iterative architectural changes to the framework so that we all have greater tools at our disposal to improve both the developer and user experience, particularly focusing on the implementation of developer tools and features that have become commonplace in other frameworks, such as Laravel.

Some of our best features are simply not finished​

There are one or two features that we see requested consistently from customers in our community forums and feedback channels, and we're excited to confirm they are coming in 2.4! However, it serves no one well if we release such highly-anticipated features before they are ready and before they have the usual level of quality, polish, and extensibility you would expect from a XenForo release. We'd rather take the extra time to get them right than rush them out and disappoint users with a subpar implementation that requires immediate patches or lacks the flexibility for customisation. We'll be sharing exciting details about what these features are and how they work in the coming weeks, so stay tuned!

We can't keep up!​

I just counted and there are about 15 features that have been merged or are pending to be merged into XF 2.4 that we haven't announced yet. Some of these are smaller and aren't worthy of a dedicated HYS of their own (so they'll probably be rolled into a "miscellaneous" HYS or two), and some of these are going to be mentioned below, but while we have been "cooking" (as the kids say these days) it has meant that things like code reviews, and writing HYS posts hasn't been easy to balance. There is also potentially more stuff coming from generous contributions from esteemed developers such as @Xon and @digitalpoint, assuming we have time to implement (otherwise they will wait for... a future version).


With all of that now being said, while 2.4 is taking longer than we wanted, we have been busy and we are very much nearing the end of development.

And, while disappointing (to all of us) it is important to maintain perspective. XenForo 2.2 was released in September 2020. XenForo 2.3 was released nearly four years later. XenForo 2.4 is not 3 more years away.

But, you clicked this to find out what's new, right? So let's go.

 

[Wildcat Media]

I tuned Cloudflare to trim away all sorts of bots and bad traffic, not just the AI slop vacuums. The biggest forum at this moment:

Total: 3,162 (members: 655, guests: 2,463), with over 36 million posts and almost 1 million threads.

That is how it traditionally was before the epidemic of bots in recent years. We usually have anywhere from ~1:2 to 1:4 ratio of members to guests, going back to 2012 when we first installed XF. When we had recent bot issues, before I spent a few days tuning and observing, we had 568 members and 11,135 guests.

 

[Mbx]

I tuned Cloudflare to trim away all sorts of bots and bad traffic, not just the AI slop vacuums. The biggest forum at this moment:

Total: 3,162 (members: 655, guests: 2,463), with over 36 million posts and almost 1 million threads.

That is how it traditionally was before the epidemic of bots in recent years. We usually have anywhere from ~1:2 to 1:4 ratio of members to guests, going back to 2012 when we first installed XF. When we had recent bot issues, before I spent a few days tuning and observing, we had 568 members and 11,135 guests.

Cloudflare helps, but your data privacy is gone with that -- to consider.

 

[ES Dev Team]

I can tell you that before the bot invasion, we had a lot of legitimate bot activity in the form of search engine scrapers.
We are heavily indexed and have good SEO.

Pre AI scraper invasion, it hovered around 1000.
This escalated to ~20,000 before the last round of fail2ban tuning.
Now it generally hangs out in the 2000-3000 range.

I am also noticing a large network in brazil. The general purpose bot protection seems to be doing a great job of batting them back, so i haven't bothered to run an analysis script to identify which ranges of IP address to ban.

The AWS Singapore network was relentless and numerous enough that i found their ranges and banned all those IPs permanently.

 

[nodle]

I am starting to think we won't see Xenforo 2.4 in the year 2025; we only have a few months left.

 

[jakellc]

Using PDO for database interactions rather than mysqli​

One of the things that we decided to do sooner, rather than later, was change our database library from mysqli to PDO. As we'll talk about later, PDO helps us write new database adapters more easily which becomes useful for supporting foreign databases for things like imports or, useful for third party developers who may wish to integrate with locally hosted foreign databases. PDO is also generally the de-facto standard for PHP at this point. Many hosts come with PDO enabled by default, whereas mysqli will often need to be enabled manually. We've seen this play out a couple of times when people change hosts or even change versions without ensuring the mysqli extension is enabled first.

To be able to install or upgrade to XenForo 2.4, you will need to ensure that PDO is available with the mysql driver in your hosting environment.

There are also some quality of life improvements included in PDO such as named parameter binding, for example:

$activeUsers = \XF::db()->fetchAll("
   SELECT *
    FROM xf_user AS user
    INNER JOIN xf_user_profile AS profile ON
        (user.user_id = profile.user_id)
    WHERE user.user_state = :user_state
        AND user.last_activity > :min_activity
    ORDER BY user.last_activity DESC
    LIMIT 10
", ['user_state' => 'valid', 'min_activity' => time()]);

Though, in retrospect, that may end up being less useful than you'd think... (more on that later).

Native support for SQLite​

SQLite is at this point the most widely deployed database in the world (across phones, embedded devices, apps etc.) with some even using it in large scale, high performance, production environments. SQLite is also great for developers due to its prevalence, zero-configuration setup, and portability (it's just a single file) with excellent performance on local disks. Despite its simplicity, SQLite is fully ACID compliant and remarkably robust. It is not just convenient, it's enterprise-grade reliable.

For advanced developers, particularly those with dev ops pipelines, SQLite lends itself extremely well to automated testing - you can spin up a complete, isolated database instance instantly for each test run. While we will still officially recommend MySQL for production, SQLite for testing and development is hugely attractive and we're excited to put it in your hands starting with XenForo 2.4.

But what about all of the existing queries in the code?

Brand new fluent query builder​

A new database means supporting a different database dialect. But no one wants to write database queries twice. In fact, it's a little bit annoying writing them once. The new query builder aims to make life easier. Let's rewrite the query I rewrote earlier, now using the new query builder:

$activeUsers = \XF::query('xf_user AS user')
    ->join('xf_user_profile AS profile', 'user.user_id', '=', 'profile.user_id')
    ->where('user.user_state', 'valid')
    ->where('user.last_activity', '>', time() - 86400)
    ->orderByDesc('user.last_activity')
    ->limit(10)
    ->fetchAll();

This query builder instance automatically compiles to the right SQL dialect for the database in use. We have rewritten all existing queries in the software (and official add-ons) to utilise the new query builder, including those generated by the Finder and the SchemaManager to support SQLite.

Of course you don't have to use it. You will be able to write different queries directly to support both MySQL (as now) and SQLite or you will be able to forget about SQLite altogether (though this may affect people's ability to use your add-on if they are using SQLite).

This also opens the door for supporting other databases in the future. While there are no immediate plans, support for MS SQL and PostgreSQL is now a possibility. Anyone who may need that right now will be able to easily write your own adapters and grammar classes for the query builder in order to add support if you need to.

A PhpStorm plugin for developers​

We acknowledge that occasionally query builders, while helpful, can sometimes slow development down because when writing raw database queries, IDEs such as PhpStorm are able to be configured to be "aware" of your database schema through its built-in database tools.

To address this issue, we're excited to announce we will be making available a free and open source PhpStorm plugin (please help us maintain it ) for PhpStorm which provides completion for table and column names.

Here it is in action:

View attachment 323730

Debugging queries​

We have implemented a few methods to help developers check what query would be compiled by any particular query builder call for debugging purposes.

$pairs = \XF::query('xf_user')
    ->dump()
    ->select('user_id', 'username')
    ->limit(5)
    ->fetchPairs();

We support dump, dd, dumpSimple, ddSimple, dumpToFile, dumpToFileNamed methods anywhere in the query builder chain. These methods register an event that fires before the final query executes, meaning it also works with aggregate methods such as count and sum.

SELECT "user_id", "username" FROM "xf_user" LIMIT 5

Speaking of debugging queries - we've also given our debug output a refreshed look, now with better colour usage in both light and dark mode and syntax highlighting and formatting for database queries:

View attachment 323723

Built-in PHP web server support​

Finding the right environment for development and testing can be challenging. There are a bunch of products available for different operating systems. Sometimes it can be difficult to find the right one for you, sometimes you might run into issues after upgrades, or you switch on your laptop one day and none of your databases are accessible. We've already talked about how SQLite can help with that - not having to manage your own MySQL or MariaDB instance is a huge benefit - but what if you didn't have to bother with Nginx or Apache, either?

# php cmd.php xf:serve
Starting XenForo development server...
XenForo development server started: http://localhost:8080
Document root is: /Users/chrisdeeming/Herd/24x
Press Ctrl-C to quit.

13:26:20 GET 200 / (432.84ms)
13:26:21 GET 200 /css.php?css=public%3Anormalize.css%2Cpublic%3Afa.css%2Cpublic%3Avariations.less%2Cpublic%3Acore.less%2Cpublic%3Aapp.less&s=1&l=1&d=1747965566&k=975f8f0746a2cbf3a7f3ee7c96f538e9a2bf61e6 (488.67ms)
13:26:21 GET 200 /css.php?css=public%3Anode_list.less%2Cpublic%3Ashare_controls.less%2Cpublic%3Aextra.less&s=1&l=1&d=1747965566&k=b21bffa3ee848ff6a7384493bd4c8b387d4a511e (58.2ms)
13:26:21 GET 200 /js/xf/preamble.js?_v=e361daff
...

Utilising the PHP built-in web server, you can quickly spin up a development server without ever having to touch web server configs or full environments like Herd, WAMP, MAMP or Docker.

XenForo Development REPL​

A new xf-dev:tinker CLI command has been added, providing a REPL (read–eval–print loop) for XenForo development. It is built on the excellent Psysh library, and provides an easy playground for running code snippets, testing services, and inspecting the environment interactively.

The REPL is preloaded with common variables, like $app, $db, $em, and many others you might (not) expect. You can view the full list with the ls -l --vars command, or disable this behavior with the --without-vars flag.

You may use the --execute option to pass code to run directly, which can be useful for running quick one-off scripts, or pass file paths as arguments to include them when the REPL starts, which can be useful for shared set-up code. The REPL also includes a --with-commands flag to allow calling other XenForo CLI commands within the REPL.

The Psysh website and docs may be helpful references for getting started with it:

• https://psysh.org/#features
• https://github.com/bobthecow/psysh/wiki/Commands

View attachment 323724

Totally understand the delay — better to take time and build a solid, future-proof 2.4 than rush it. Excited to see what’s coming!

 

[smallwheels]

I am starting to think we won't see Xenforo 2.4 in the year 2025; we only have a few months left.

Maybe they should rename 2.4 to "XenForo Forever" as a way of expectation management including the possibility to surprise us in a positive way by an unexpected earlier release?

(...) Duke Nukem Forever began development under 3D Realms and underwent a severely protracted development that lasted 14 years. Announced in 1997 following the critical and commercial success of Duke Nukem 3D, it was delayed several times, which was attributed to engine changes, understaffing, and a lack of a development plan. After 3D Realms downsized in 2009, Duke Nukem Forever was finished by Triptych Games, Gearbox Software, and Piranha Games. It holds the Guinness world record for the longest development for a video game.

Duke Nukem Forever was released on June 14, 2011 (...)

 

[ES Dev Team]

Current mood on that topic:

 

[Goodfella]

AI successfully replaced the Froala editor with the TipTap editor in just 10 minutes. It’s impressive to imagine what it could accomplish in a few hours, days, weeks, or even months.

 

[ES Dev Team]

And does it pass 100% of your expected functionality tests, including handling uploaded images and media embeds?

How is converting content handled?

 

[Goodfella]

And does it pass 100% of your expected functionality tests, including handling uploaded images and media embeds?

How is converting content handled?

I can confirm that image attachments are working — they upload successfully and attach to the post. I haven’t explored it in depth yet; as mentioned, it only took about ten minutes of prompting the AI model. I’m confident that with more time, it could be refined to work flawlessly. This was mainly a proof of concept to demonstrate that development shouldn’t take this long given the tools available today.

 

[ES Dev Team]

I'd call it a success when you look at it in depth, fix whatever problems exist ( possibly by hand ), and know that you have all the other small details handled. Time yourself too!

You are having AI do the easy part and i'm betting a number of things would have to be done by hand to get a complete implementation, which is time consuming.

 

[Jimsef]

Time consuming, weeks, yes… time consuming 6 months to a year… errr

 

[RobinHood]

Well, Cursor is currently down, so I guess we're all back to the good old days of thinking for ourselves and searching stack overflow!

Ahhhh, who am I kidding!

Time to pack it in for the night and hope it's all working again in the morning!

 

[🔥Iggy🔥]

When we announced that Tiptap is coming to XenForo 2.4 it was 95% complete, and we then took a bit of a pause to work on other projects, which we have talked about since and will be discussing in this thread. Since then, Tiptap have announced Tiptap V3 which is currently in beta. Given how core the editor is to the forum experience, it makes a lot of sense to ship XenForo 2.4 with Tiptap V3 rather than Tiptap V2 as originally planned. While the changes involved are not too extensive, we also don't want to ship 2.4 with a dependency that is still in beta and subject to change. While we are not planning to wait for Tiptap V3 to be stable, necessarily, we do at least want to give it a little bit more time so we have a higher degree of confidence that we're shipping a stable editing experience.

this thread made perfect sense when it was posted.....enough time has passed however, its appropriate to update xenforo customers again....much of this really isnt valid anymore.....tiptap3 is well out of beta its at version 3.7.2 now......minor features you were finishing up etc.......is this about the code refactor comic? you are distressing ppl bruh
maybe people are different but i think its much better to hear a bad news than silence.....unless its really just so bad you are afraid to say......

 

[ES Dev Team]

I have a 5090 here serving a LLM 90% as good as Deepseek R1 ( SEED-OSS 36B ) and it only goes down when i press the off button.
$2500 USD + electricity = a lifetime of predictable service

I do a little vibe coding but even the state of the art is not up to complex programs yet, and LLM code does not have good taste and is a pain to work with later, so it can't be used for many serious things.

No freaking way would i trust even Claude to do any serious work without implementing some tech debt for you to trip over later.
Good short term move, really bad long term move when you find yourself rewriting a bunch of code.

As a customer i want to purchase software that was crafted thoughfully, not artificially thoughtfully

 

[Jimsef]

I have a 5090 here serving a LLM 90% as good as Deepseek R1 ( SEED-OSS 36B ) and it only goes down when i press the off button.
$2500 USD + electricity = a lifetime of predictable service

I do a little vibe coding but even the state of the art is not up to complex programs yet, and LLM code does not have good taste and is a pain to work with later, so it can't be used for many serious things.

No freaking way would i trust even Claude to do any serious work without implementing some tech debt for you to trip over later.
Good short term move, really bad long term move when you find yourself rewriting a bunch of code.

As a customer i want to purchase software that was crafted thoughfully, not artificially thoughtfully

That is hilarious... most of that post was just nonsense.

Clearly the previous poster was expressing a frustration that you can achieve a lot in a short timeframe using AI... and a solid dev team could achieve even more. I've worked in software development for over a decade and I've never seen such a mess of communication and lack of progress. This just feels like a hobbyist organisation pretending to be a professional one.

Where are the timelines, where is the roadmap? Both things that commercial organisations build their reputation around.

 

[mjda]

Where are the timelines, where is the roadmap? Both things that commercial organisations build their reputation around.

Isn't it obvious? The roadmap is in the HYS threads. The timeline is "coming soon", as seen here: https://xenforo.com/community/threads/coming-soon-xenforo-2-4.225302/. The reputation speaks for itself...just look at the replies in this thread!

On a more serious note, though, XF has never given a roadmap or strict timelines, at least not that I've seen. When they do, it's things like this, which do more harm than good when you can't meet those newly created expectations. I don't think most people would worry too much about roadmaps or timelines if it even remotely looked like the devs, or other staff, were still even interested in this place. The communication has definitely gone by the wayside which is, to me, the biggest issue that's going on here. That certainly could be because there simply isn't anything to communicate, but I'd argue just saying that every so often would calm a lot of this negativity. It'd be much better than the near radio silence that we're getting now.

 

[Wing]

Isn't it obvious? The roadmap is in the HYS threads. The timeline is "coming soon", as seen here: https://xenforo.com/community/threads/coming-soon-xenforo-2-4.225302/. The reputation speaks for itself...just look at the replies in this thread!

On a more serious note, though, XF has never given a roadmap or strict timelines, at least not that I've seen. When they do, it's things like this, which do more harm than good when you can't meet those newly created expectations. I don't think most people would worry too much about roadmaps or timelines if it even remotely looked like the devs, or other staff, were still even interested in this place. The communication has definitely gone by the wayside which is, to me, the biggest issue that's going on here. That certainly could be because there simply isn't anything to communicate, but I'd argue just saying that every so often would calm a lot of this negativity. It'd be much better than the near radio silence that we're getting now.

Just 4 months ago, you and others in here were arguing with some of us that were being realists and saying exactly what you yourself now are saying. What happened?

 

[mjda]

Just 4 months ago, you and others in here were arguing with some of us that were being realists and saying exactly what you yourself now are saying. What happened?

That's not true. You were crying about the lack of software updates and singing gloom and doom for XF as a company. I simply said, like I've always said, that the lack of communication is my primary issue with XF. It's not important to me how long updates take to come out, or if they even come out at all. I don't rely on XF to keep my sites relevant. My stance hasn't changed.

 

[beerForo]