perf: avoid unnecessary clone

[tsctx]

Deps: #3109

Optimize cases where the body is a string, BufferSource, URLSearchParams, or stream (only one chunk).

Avoid cloning if unnecessary.

• perf(fetch): improve body mixin methods #3109

benchmark                 time (avg)             (min … max)       p75       p99      p999
------------------------------------------------------------ -----------------------------
• small (length 512)
------------------------------------------------------------ -----------------------------
Response#arrayBuffer  22'113 ns/iter  (11'200 ns … 3'439 µs) 19'100 ns    116 µs    441 µs
Response#text         16'603 ns/iter    (11'700 ns … 673 µs) 15'100 ns 51'600 ns    268 µs

summary for small (length 512)
  Response#text
   1.33x faster than Response#arrayBuffer

• middle (length 8192)
------------------------------------------------------------ -----------------------------
Response#arrayBuffer  25'506 ns/iter  (12'500 ns … 3'582 µs) 21'600 ns    113 µs    337 µs
Response#text         27'878 ns/iter    (14'000 ns … 885 µs) 25'900 ns    115 µs    321 µs

summary for middle (length 8192)
  Response#arrayBuffer
   1.09x faster than Response#text

• long (length 131072)
------------------------------------------------------------ -----------------------------
Response#arrayBuffer     233 µs/iter  (40'300 ns … 3'191 µs)    218 µs  1'369 µs  2'948 µs
Response#text            374 µs/iter     (137 µs … 3'573 µs)    350 µs  2'034 µs  3'573 µs

summary for long (length 131072)
  Response#arrayBuffer
   1.61x faster than Response#text

• this patch

benchmark                 time (avg)             (min … max)       p75       p99      p999
------------------------------------------------------------ -----------------------------
• small (length 512)
------------------------------------------------------------ -----------------------------
Response#arrayBuffer  22'036 ns/iter  (11'300 ns … 2'884 µs) 21'100 ns    106 µs    453 µs
Response#text         13'975 ns/iter  (10'200 ns … 1'246 µs) 13'200 ns 37'900 ns    235 µs

summary for small (length 512)
  Response#text
   1.58x faster than Response#arrayBuffer

• middle (length 8192)
------------------------------------------------------------ -----------------------------
Response#arrayBuffer  25'928 ns/iter  (12'600 ns … 4'707 µs) 22'400 ns    111 µs    351 µs
Response#text         18'282 ns/iter    (11'900 ns … 473 µs) 17'400 ns 88'400 ns    291 µs

summary for middle (length 8192)
  Response#text
   1.42x faster than Response#arrayBuffer

• long (length 131072)
------------------------------------------------------------ -----------------------------
Response#arrayBuffer     221 µs/iter  (39'700 ns … 3'282 µs)    206 µs    872 µs  3'059 µs
Response#text            265 µs/iter     (124 µs … 2'542 µs)    247 µs  1'394 µs  2'504 µs

summary for long (length 131072)
  Response#arrayBuffer
   1.2x faster than Response#text

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 94.06%. Comparing base (7a94682) to head (ac0787f).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #3117   +/-   ##
=======================================
  Coverage   94.06%   94.06%           
=======================================
  Files          89       89           
  Lines       24327    24336    +9     
=======================================
+ Hits        22883    22892    +9     
  Misses       1444     1444           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[mcollina]

lgtm