Open-source project
TruCopilot/phpfastcache avatar
TruCopilot/phpfastcache

phpfastcache's readme points fifteen links at a different organisation, five drivers carry a deprecation notice, and its release line went quiet for seventeen months

A high-performance backend cache system. It is intended for use in speeding up dynamic web applications by alleviating database load. Well implemented, it can drops the database load to almost nothing, yielding faster page load times for users, better resource utilization. It is simple yet powerful.

2,408 stars441 forksPHPMIT

At a glance

What is it?
TruCopilot/phpfastcache is a multi-backend cache library for PHP whose page documents a nine-to-ten migration that replaced the configuration array with an object, whose driver table is mid-way through moving six backends out of core into separate composer packages, and whose newest release sits seventeen months behind the one before it.
Who is it for?
This is worth reading if you need one cache interface across many backends and want to plan an upgrade, because the deprecation and extraction notes on the page are unusually specific about what is going away and when, and the migration guide it points at is the thing you will actually need. Two things to check first.
Can I use it commercially?
Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
Is it still maintained?
Activity is slowing. The repository last received commits 7 months ago.
What is it written in?
Mainly PHP, according to GitHub's language statistics.

Answers come from the project's GitHub data, last synced on November 15, 2026, and from our analysis. They are not legal advice.

From daily snapshots

Star history

  • TruCopilot/phpfastcache2,408 ★ · -2
Star history
Repository2026-09-302026-11-15
TruCopilot/phpfastcache2.4k2.4k

-2 Stars in the last 46 days, measured from the daily GitHub snapshot. Compare with other projects

Editorial analysis

Every internal link points at a different organisation

The repository sits under one name and the readme points everywhere else.

The badges, the driver extension sub-repositories, the wiki link, the link inviting pull requests to propose a new driver, the code quality badges and the test workflow link all resolve through a different organisation's path. Counting them, there are fifteen such references in the page and not one occurrence of the name this repository is published under.

The package index links point at a third name, which is the conventional one for the package rather than the repository. So there are three identities in play: the repository name, the package name on the index, and the organisation every internal link still uses.

Two smaller signals in the same direction. The funding link in the badge row goes to an individual's page rather than to an organisation. And the licence file at the repository root is spelled with a c, which is the older convention and will not be found by a tool looking for the usual spelling.

None of this affects the code. It affects the reader, who follows a driver extension link expecting it to be maintained alongside this repository and lands somewhere else.

Five drivers carry a deprecation note and one is already gone

The driver table is four columns wide and the maintenance state is written into the cells rather than into a separate note, which means you only see it by reading across the table.

One regular driver has already lost its dependency: the opcode cache driver is listed with a note that support was removed, and the row is marked as core, so the slot is reserved rather than deleted.

One more is deprecated as of the current minor version with a note that it will be removed in the next major version. That is the Windows cache driver, and the version numbers in the note are the ones to write down.

In the high performance column there are three more. One is marked as a version that will be deprecated as of the next major, which is the older Couchbase client. One is the Cassandra driver, with a parenthetical explaining that the underlying extension is no longer maintained upstream by its vendor and that the driver might therefore be deprecated in the next major. And one is a caching relay service whose support the page ties to a date at the end of 2024.

So five of the listed backends carry a forward-looking warning, and in three of those cases the reason is upstream rather than the library's own plan. That is a more useful categorisation than a bare deprecation list would be.

Six backends moved out of core into their own composer packages

The structural change underneath the driver table is that the core has been getting smaller.

Six backends are named as having been moved from the core into standalone extension repositories: two document databases, two NoSQL stores, one document store with sub-queries and one search engine. They are added through the package manager rather than shipped with the library, and the page gives the shape of the command.

That is a defensible move for optional dependencies that pull their own client libraries, and it makes the core install smaller. The cost is that a feature you previously got by changing a configuration string now needs a second package.

One backend is the exception and the page is explicit about it: the older Couchbase client stays in the core for compatibility reasons, and is marked as deprecated. So there are two Couchbase drivers in the table, one in core and deprecated and one as a separate extension, and the naming differs only by a version suffix. That is a migration hazard rather than a feature.

The same pattern appears in the driver list's other columns. Two development-only drivers are marked as core, one of which is an in-memory driver listed with a note that it was previously named something else, so the rename is recorded next to the row it affects.

Version 9 replaced the configuration array with an object

The migration notice at the top of the page is the most consequential thing in it, and it is three sentences long.

The first is a warning that the current major line is mostly a move to type-aware coding for the language with some significant changes, and that it is relatively not compatible with earlier versions. The word relatively is doing work in that sentence.

The second names the biggest change: the configuration system is now an object, where it used to be a primitive array. That is a breaking change in the shape of the object every consumer touches, and it is the reason the page sends you to a migration guide before anything else.

The third is a floor, stated with a grammatical slip: the current line requires at least the newer major version of the language to work properly.

So adopting version 9 is not a version bump. It is a configuration rewrite with a guide, on a library whose own page describes the compatibility as partial. A codebase with a large configuration array has to do that work before it can take the upgrade, and the guide is the only documentation for it.

The minor releases within the current line, by contrast, look conservative, which is a reasonable strategy for a library whose major version already absorbed the breaking change.

The release line went quiet for seventeen months

There are three releases visible and the gaps between them are the story.

Two of them landed three days apart in January 2024. The third came seventeen months later, in April 2025. The default branch is not archived, and its most recent push is dated 2026-04-07, about six months after the newest release.

So the published version has been unchanged since April 2025 while the branch has continued to receive work. A page that documents deprecations for the next major version is describing a release that does not exist yet.

There is a documentation detail worth pairing with that. Two of the three releases carry no release note at all in the list; only the newest one has a title beyond the version number. The repository does have two change log files at its root, one for the library and one for its public interface, so the history exists in the tree even where the release list does not carry it.

The page also describes the project as actively developing one specific extension, which is narrower than saying the library is actively developed. Given the dates above, the narrower claim is the one to rely on.

Two analyser configurations and a Windows batch file for quality

The repository root of a caching library carries a broad quality toolchain, and one pair of entries in it is unusual.

There is a standard static analyser configuration, and there is a second one beside it whose name suggests a reduced or relaxed mode of the same analyser. Two configurations for one tool means either a fast subset for pre-commit and a full pass for CI, or a graduated adoption, and the page does not say which.

Around them are a coding-standards configuration and a separate dead-code and complexity configuration, two changelog files, a security policy, a coding guideline document, a document listing the extension packages, and a custom domain file for the project pages.

The quality tooling is driven by a Windows batch file at the root. So the whole quality suite is invoked from a script that will not run on anything but one operating system, which is an odd choice for a library that lists cross-platform portability as one of its three goals.

That goal itself is one of three the page states: performance, security, and portability. The security one is stated with an unusual pair of words, describing a library that is both safe and strong, on the grounds that caching strategies can introduce vulnerabilities. For a library whose core abstraction lets one cache key shape reach every backend, that is the right thing to worry about.

The abstraction is one class for many backends, and the page leans on it

The case the page makes is narrow and simple: one class for many backend caches, so you do not rewrite your code when you change where the cache lives.

That is the whole product, and everything else in the driver table is evidence for it. Local memory, opcode cache, files, a local SQL database, a local key-value store, a distributed one, and four replication topologies ranging from a master with slaves to full replication, plus two development drivers that do nothing and two in-process development stores.

The four replication topologies are the interesting corner of that list, because they are marked as core rather than as extensions, which means clustering is part of the abstraction rather than an add-on. Their names describe the strategy directly: one topology propagates every write to every node, one propagates to a subset, and the remaining two sit between full propagation and a single writable node.

A predictable one is in the same column, which is the strategy for when any node may fail and the read can go anywhere.

The page's framing is that caching should not mean weakening your code, and the three goals behind that are performance, security and portability. The abstraction is what delivers the portability half, and it is also what makes the security half harder, since one cache-key shape has to be correct for every backend.

Editorial conclusion

This is worth reading if you need one cache interface across many backends and want to plan an upgrade, because the deprecation and extraction notes on the page are unusually specific about what is going away and when, and the migration guide it points at is the thing you will actually need. Two things to check first. The readme's own links still resolve through a different organisation, so anything you follow from it may not track this repository. And version 9 changed the configuration type, so code written against the array form will not run against the object form without going through the documented migration.

Frequently asked questions

What is phpfastcache and what changed in version 9?

It is a PHP cache library with one abstraction over many backends, from local memory and files to Redis, clustered replication topologies and document databases. Version 9's biggest change is that the configuration is now an object rather than a primitive array, which the page describes as only relatively compatible with earlier versions.

Which phpfastcache drivers are deprecated?

Five carry a warning. The opcode cache driver has already lost its support and is marked as removed, the Windows cache driver is deprecated as of the current minor version, and the older Couchbase client, the Cassandra driver and a caching relay service are all marked as deprecated or possibly deprecated in the next major version.

Do I need a separate package to use some phpfastcache backends?

Yes for six of them. Two document databases, two NoSQL stores, a document store with sub-queries and a search engine have been moved out of the core into standalone extension repositories that are added through the package manager. The older Couchbase client stays in the core for compatibility but is deprecated.

What are the newest phpfastcache releases?

Version 9.2.4 from April 2025, after 9.2.3 from January 2024 and 9.2.2 two days earlier. The gap between the two January releases and the April one is about seventeen months, and the default branch was last pushed to on 2026-04-07 without a release since.

Does phpfastcache cache clustering?

Yes, and the cluster topologies are part of the core rather than an extension. The table lists full replication, semi replication, master with slaves, and a random-replication strategy for when any node may fail and reads can go anywhere.

Official sources

  1. License: MIT
  2. Project website
  3. README
  4. Releases
  5. TruCopilot/phpfastcache on GitHub
Add this badge to your README

If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.

Add this badge to your README

markdown
[![Hysen Labs](https://hysenlabs.com/badge/trucopilot-phpfastcache.svg)](https://hysenlabs.com/projects/trucopilot-phpfastcache)

Star history chart for your README

markdown
[![Star history](https://hysenlabs.com/chart/trucopilot-phpfastcache.svg)](https://hysenlabs.com/projects/trucopilot-phpfastcache)

Was this useful?