Slim vs Haml

slim_vs_haml.md

Analysis of Slim vs. Haml Project Health

• Static data as of April 13, 2015, some updates as of October 1, 2015

#	Metric	Haml	Slim	Winner
1	Issues			Slim
2	Stars			Slim
3	Quality		--	Haml
4	Test Coverage		--	Haml
5	Documentation			Slim
6	Watchers	95	112	Slim
7	Forks	440	306	Haml
8	Pull Requests	11	1	Slim
9	Activity	1 commmit / Month	6 commits / Month	Slim
10	Community	1 author / Month	4 authors / Month	Slim
11	Raw HTML Compat	No	Yes	Slim
12	Rails Streaming Compat	No, and issue was closed in 2020	Yes	Slim
13	Speed*	1.09x slower than ERB	1.12x slower than ERB	-

*References

Rails Streaming Compatibility

The Haml issue has open since Sep 23, 2011, was bumped from 4.0 release, to 5.0 release, now to 6.0 release, with no end in sight, and people openly quesitoning if it is even possible. Haml has a notoriously difficult codebase.

Speed

Conventional wisdom was that Haml is slow, Slim is fast. However haml was rewritten to use hamlit based on a better parser and compiler optimizations (Ripper and Temple). Now haml is as fast as, or faster than slim. They are now roughly equivalent in speed, meaning speed is no longer a deciding factor between them.

In 2022 v6 of Haml was refreshed using the internals of Hamlit, and announced that it was finally faster than slim. Slim in turn was jolted back to life, releasing v5, and in 2023 slim may once again be faster.

slim-template/benchmarks#4

In February 2024 the various libraries tested in this benchmark are nearly identical in a basic performance test. Of course, YMMV depending on which features you use, and how much.

Test was run on a 16GB 13" M2 MacBook Pro running MacOS Sonoma 14.3.1.

❯ bundle install
Fetching gem metadata from https://rubygems.org/..
Resolving dependencies...
Fetching temple 0.10.3
Installing temple 0.10.3
Fetching slim 5.2.1
Fetching haml 6.3.0
Installing slim 5.2.1
Installing haml 6.3.0
Bundle complete! 6 Gemfile dependencies, 8 gems now installed.
Use `bundle info [gemname]` to see where a bundled gem is installed.
❯ ./run-benchmarks.rb
ruby 3.2.2 (2023-03-30 revision e51014f9c0) [arm64-darwin22]
Warming up --------------------------------------
           erb 4.0.2    61.264k i/100ms
        erubi 1.12.0    64.087k i/100ms
   temple erb 0.10.3    58.292k i/100ms
          haml 6.3.0    60.975k i/100ms
          slim 5.2.1    60.300k i/100ms
Calculating -------------------------------------
           erb 4.0.2    656.704k (± 2.4%) i/s -      3.308M in   5.040677s
        erubi 1.12.0    670.228k (± 0.9%) i/s -      3.397M in   5.068270s
   temple erb 0.10.3    581.279k (± 1.0%) i/s -      2.915M in   5.014608s
          haml 6.3.0    612.146k (± 0.4%) i/s -      3.110M in   5.080135s
          slim 5.2.1    596.732k (± 6.4%) i/s -      3.015M in   5.088974s

Comparison:
        erubi 1.12.0:   670227.5 i/s
           erb 4.0.2:   656703.7 i/s - same-ish: difference falls within error
          haml 6.3.0:   612145.8 i/s - 1.09x  slower
          slim 5.2.1:   596731.6 i/s - 1.12x  slower
   temple erb 0.10.3:   581278.6 i/s - 1.15x  slower

./run-benchmarks.rb  35.35s user 0.35s system 97% cpu 36.554 total

Performance is no longer the reason to choose Slim!