text-builder-linear

Linear types for linear times!

Builder for strict Text and ByteString, based on linear types. It consistently outperforms lazy Builder from text as well as a strict builder from text-builder, and scales better.

Example

> :set -XOverloadedStrings
> import Data.Text.Builder.Linear
> fromText "foo" <> fromChar '_' <> fromDec (42 :: Int)
"foo_42"

Design

String builders in Haskell serve the same purpose as StringBuilder in Java to prevent quadratic slow down in concatenation.

Classic builders such as Data.Text.Lazy.Builder are lazy and fundamentally are dlist with bells and whistles: instead of actually concatenating substrings we compose actions, which implement concatenation, building a tree of thunks. The tree can be forced partially, left-to-right, producing chunks of strict Text, combined into a lazy one. Neither input, nor output need to be materialized in full, which potentially allows for fusion. Such builders allow linear time complexity, but constant factors are relatively high, because thunks are expensive. To a certain degree this is mitigated by inlining, which massively reduces number of nodes.

Strict builders such as text-builder offer another design: they first inspect their input in full to determine output length, then allocate a buffer of required size and fill it in one go. If everything inlines nicely, the length may be known in compile time, which gives blazingly fast runtime. In more complex cases it still builds a tree of thunks and forces all inputs to be materialized.

This package offers two interfaces. One is a mutable Buffer with linear API, which operates very similar to StringBuilder in Java. It allocates a buffer with extra space at the ends to append new strings. If there is not enough free space to insert new data, it allocates a twice larger buffer and copies itself there. The dispatch happens in runtime, so we do not need to inspect and materialize all inputs beforehand; and inlining is mostly irrelevant. Exponential growth provides for amortized linear time. Such structure can be implemented without linear types, but that would greatly affect user experience by polluting everything with ST monad. Users are encouraged to use Buffer API, and built-in benchmarks refer to it.

The second interface is more traditional newtype Builder = Builder (Buffer ⊸ Buffer) with Monoid instance. This type provides easy migration from other builders, but may suffer from insufficient inlining, allocating a tree of thunks. It is still significantly faster than Data.Text.Lazy.Builder, as witnessed by benchmarks for blaze-builder below.

Case study

Let's benchmark builders, which concatenate all Char from minBound to maxBound, producing a large Text:

#!/usr/bin/env cabal
{- cabal:
build-depends: base, tasty-bench, text, text-builder, text-builder-linear
ghc-options: -O2
-}

import qualified Data.Text as T
import qualified Data.Text.Lazy as TL
import qualified Data.Text.Lazy.Builder as TLB
import qualified Text.Builder as TB
import qualified Data.Text.Builder.Linear as TBL
import System.Environment (getArgs)
import Test.Tasty.Bench

mkBench :: Monoid a => String -> (Char -> a) -> (a -> Int) -> Benchmark
mkBench s f g = bench s $ nf (g . foldMap f . enumFromTo minBound) maxBound
{-# INLINE mkBench #-}

main :: IO ()
main = defaultMain
  [ mkBench "text, lazy" TLB.singleton (fromIntegral . TL.length . TLB.toLazyText)
  , mkBench "text, strict" TLB.singleton (T.length . TL.toStrict . TLB.toLazyText)
  , mkBench "text-builder" TB.char (T.length . TB.run)
  , mkBench "text-builder-linear" TBL.fromChar (T.length . TBL.runBuilder)
  ]

Running this program with cabal run Main.hs -- +RTS -T yields following results:

text, lazy:
  4.25 ms ± 107 μs,  11 MB allocated, 912 B  copied
text, strict:
  7.18 ms ± 235 μs,  24 MB allocated,  10 MB copied
text-builder:
  80.1 ms ± 3.0 ms, 218 MB allocated, 107 MB copied
text-builder-linear:
  5.37 ms ± 146 μs,  44 MB allocated,  78 KB copied

The first result seems the best both in time and memory and corresponds to the usual Text builder, where we do not materialize the entire result at all. It builds chunks of lazy Text lazily and consumes them at once by TL.length. Thus there are 11 MB of allocations in nursery, none of which survive generation 0 garbage collector, so nothing is copied.

The second result is again the usual Text builder, but emulates a strict consumer: we materialize a strict Text before computing length. Allocation are doubled, and half of them (corresponding to the strict Text) survive to the heap. Time is also almost twice longer, but still quite good.

The third result is for text-builder and demonstrates how bad things could go with strict builders, aiming to precompute the precise length of the buffer: allocating a thunk per char is tremendously slow and expensive.

The last result corresponds to the current package. We generate a strict Text by growing and reallocating the buffer, thus allocations are quite high. Nevertheless, it is already faster than the usual Text builder with strict consumer and does not strain the garbage collector.

Things get very different if we remove {-# INLINE mkBench #-}:

text, lazy:
  36.9 ms ± 599 μs, 275 MB allocated,  30 KB copied
text, strict:
  44.7 ms ± 1.3 ms, 287 MB allocated,  25 MB copied
text-builder:
  77.6 ms ± 2.2 ms, 218 MB allocated, 107 MB copied
text-builder-linear:
  5.35 ms ± 212 μs,  44 MB allocated,  79 KB copied

Builders from text package degrade rapidly, 6-8x slower and 10-20x more allocations. That's because their constant factors rely crucially on everything getting inlined, which makes their performance fragile and unreliable in large-scale applications. On the bright side of things, our builder remains as fast as before and now is a clear champion.

Benchmarks for Text

Measured with GHC 9.6 on aarch64:

Group / size	text	text-builder		This package
Text
1	47.4 ns	24.2 ns	0.51x	35.2 ns	0.74x
10	509 ns	195 ns	0.38x	197 ns	0.39x
100	4.94 μs	1.74 μs	0.35x	1.66 μs	0.34x
1000	52.6 μs	17.0 μs	0.32x	15.0 μs	0.28x
10000	646 μs	206 μs	0.32x	155 μs	0.24x
100000	12.2 ms	3.34 ms	0.27x	2.60 ms	0.21x
1000000	159 ms	55.3 ms	0.35x	16.1 ms	0.10x
Char
1	46.9 ns	21.1 ns	0.45x	22.3 ns	0.48x
10	229 ns	152 ns	0.66x	79.9 ns	0.35x
100	2.00 μs	1.23 μs	0.61x	618 ns	0.31x
1000	21.9 μs	10.3 μs	0.47x	6.28 μs	0.29x
10000	285 μs	153 μs	0.54x	68.5 μs	0.24x
100000	7.70 ms	4.08 ms	0.53x	992 μs	0.13x
1000000	110 ms	106 ms	0.96x	9.19 ms	0.08x
Decimal
1	97.7 ns	872 ns	8.92x	80.2 ns	0.82x
10	864 ns	8.72 μs	10.09x	684 ns	0.79x
100	9.07 μs	93.5 μs	10.32x	7.25 μs	0.80x
1000	92.4 μs	1.06 ms	11.44x	67.5 μs	0.73x
10000	1.13 ms	13.4 ms	11.88x	667 μs	0.59x
100000	18.7 ms	141 ms	7.57x	7.57 ms	0.41x
1000000	229 ms	1.487 s	6.48x	67.8 ms	0.30x
Hexadecimal
1	403 ns	749 ns	1.86x	43.9 ns	0.11x
10	3.94 μs	7.66 μs	1.94x	308 ns	0.08x
100	42.8 μs	89.0 μs	2.08x	2.88 μs	0.07x
1000	486 μs	986 μs	2.03x	27.7 μs	0.06x
10000	7.10 ms	12.6 ms	1.77x	283 μs	0.04x
100000	80.1 ms	133 ms	1.65x	3.53 ms	0.04x
1000000	867 ms	1.340 s	1.55x	28.9 ms	0.03x
Double
1	7.56 μs	18.3 μs	2.42x	414 ns	0.05x
10	76.5 μs	188 μs	2.46x	4.23 μs	0.06x
100	754 μs	2.35 ms	3.11x	44.4 μs	0.06x
1000	7.94 ms	25.8 ms	3.25x	436 μs	0.05x
10000	79.1 ms	285 ms	3.60x	4.90 ms	0.06x
100000	796 ms	2.938 s	3.69x	45.1 ms	0.06x
1000000	8.003 s	32.411 s	4.05x	436 ms	0.05x

If you are not convinced by synthetic data, here are benchmarks for blaze-markup after migration to Data.Text.Builder.Linear:

bigTable
  992  μs ±  80 μs, 49% less than baseline
basic
  4.35 μs ± 376 ns, 47% less than baseline
wideTree
  1.26 ms ±  85 μs, 53% less than baseline
wideTreeEscaping
  217  μs ± 7.8 μs, 58% less than baseline
deepTree
  242  μs ±  23 μs, 48% less than baseline
manyAttributes
  811  μs ±  79 μs, 58% less than baseline
customAttribute
  1.68 ms ± 135 μs, 56% less than baseline

Benchmarks for ByteString

Somewhat surprisingly, text-builder-linear now offers rendering to strict ByteString as well. It is consistently faster than bytestring when a string gets over 32k (which is defaultChunkSize for bytestring builder). For mid-sized strings bytestring is slightly faster in certain disciplines, mostly by virtue of using cbits via FFI, while this package remains 100% native Haskell.

Benchmarks below were measured with GHC 9.6 on aarch64 and include comparison to bytestring-strict-builder:

Group / size	bytestring	…-strict-builder		This package
Text
1	106 ns	33.5 ns	0.32x	35.2 ns	0.33x
10	322 ns	217 ns	0.68x	197 ns	0.61x
100	2.49 μs	1.89 μs	0.76x	1.66 μs	0.67x
1000	21.8 μs	18.5 μs	0.85x	15.0 μs	0.69x
10000	231 μs	212 μs	0.92x	155 μs	0.67x
100000	3.97 ms	3.54 ms	0.89x	2.60 ms	0.66x
1000000	81.2 ms	51.5 ms	0.63x	16.1 ms	0.20x
Char
1	99.0 ns	19.4 ns	0.20x	22.3 ns	0.23x
10	270 ns	82.9 ns	0.31x	79.9 ns	0.30x
100	1.77 μs	723 ns	0.41x	618 ns	0.35x
1000	20.4 μs	8.37 μs	0.41x	6.28 μs	0.31x
10000	322 μs	129 μs	0.40x	68.5 μs	0.21x
100000	10.4 ms	2.50 ms	0.24x	992 μs	0.10x
1000000	143 ms	67.4 ms	0.47x	9.19 ms	0.06x
Decimal
1	152 ns	174 ns	1.14x	80.2 ns	0.53x
10	685 ns	1.55 μs	2.26x	684 ns	1.00x
100	5.88 μs	17.2 μs	2.93x	7.25 μs	1.23x
1000	60.3 μs	196 μs	3.25x	67.5 μs	1.12x
10000	648 μs	4.25 ms	6.57x	667 μs	1.03x
100000	11.2 ms	62.8 ms	5.62x	7.57 ms	0.68x
1000000	150 ms	655 ms	4.37x	67.8 ms	0.45x
Hexadecimal
1	94.7 ns			43.9 ns	0.46x
10	255 ns			308 ns	1.21x
100	1.72 μs			2.88 μs	1.67x
1000	18.9 μs			27.7 μs	1.46x
10000	250 μs			283 μs	1.13x
100000	6.94 ms			3.53 ms	0.51x
1000000	93.2 ms			28.9 ms	0.31x
Double
1	457 ns			414 ns	0.91x
10	3.94 μs			4.23 μs	1.07x
100	40.3 μs			44.4 μs	1.10x
1000	398 μs			436 μs	1.10x
10000	5.65 ms			4.90 ms	0.87x
100000	63.3 ms			45.1 ms	0.71x
1000000	673 ms			436 ms	0.65x