When loading the @iconify/tailwind4 plugin, it causes unexpected CPU and memory usage. Below are the details.
details
Windows 11 25H2 (reproduced by @HowieHz)
(To prevent the computer from freezing, I interrupted the formatting process.)
Compared to when this configuration option is not set.
Ubuntu 26.04.03 LTS (reproduced by @xingwangzhe)

By the way, this also leads to longer execution times in GitHub CI than Prettier.
Upon investigation, it was found that loading the @plugin "@iconify/tailwind4" plugin like this:
@plugin "@iconify/tailwind4" {
prefixes:
mdi, heroicons-outline, fa-solid, fluent, material-symbols, ic, typcn, simple-icons, ant-design, ri, ph,
icomoon-free, entypo-social, uil, cib, tabler, pepicons-pencil;
}
causes CPU usage to max out, even without any prefixes. Then, as more prefixes are added, memory usage grows at an alarming rate.
The image below shows the memory usage when prefixes are added one by one.
Steps to Reproduce
# The default branch has set --threads 2 to avoid memory exhaustion.
# You need to go to package.json and remove this configuration.
git clone https://github.com/HowieHz/halo-theme-higan-hz
pnpm i
pnpm fmt:oxfmt
I've created a similar but smaller reproduction case.
git clone https://github.com/HowieHz/test-oxfmt-bug
pnpm i
pnpm fmt:oxfmt
When loading the @iconify/tailwind4 plugin, it causes unexpected CPU and memory usage. Below are the details.
details
Windows 11 25H2 (reproduced by @HowieHz)
(To prevent the computer from freezing, I interrupted the formatting process.)
Compared to when this configuration option is not set.
Ubuntu 26.04.03 LTS (reproduced by @xingwangzhe)

By the way, this also leads to longer execution times in GitHub CI than Prettier.
Upon investigation, it was found that loading the
@plugin "@iconify/tailwind4"plugin like this:causes CPU usage to max out, even without any prefixes. Then, as more prefixes are added, memory usage grows at an alarming rate.
The image below shows the memory usage when prefixes are added one by one.
Steps to Reproduce
I've created a similar but smaller reproduction case.