panates / postgrejs / 34691284673

98%

master: 89%

push

github

refactor: build SmartBuffer/BufferReader on top of flexy-buffer

SmartBuffer and BufferReader now extend flexy-buffer's FlexyBuffer and
BufferReader instead of hand-rolling the same growable-buffer/cursor
logic, adding only the PostgreSQL wire format's two string encodings
(NUL-terminated C strings, length-prefixed strings with a negative
length as the NULL sentinel) on top.

This changes the write methods' return value from `this` to the number
of bytes written (flexy-buffer's convention, and arguably the more
useful one for length-patching code), and renames offset/length/
readBuffer/writeBuffer to flexy-buffer's position/size/readBytes/
writeBytes throughout - so every call site in frontend.ts, backend.ts,
pgoutput.ts, and the binary array/jsonb helpers moves off the fluent
chaining style and the old names.

Also drops SmartBuffer's manual writeBigInt64BE fallback for Node
builds without a native Buffer.writeBigInt64BE, now genuinely dead
code given the project's supported Node range (and already assumed
available elsewhere, e.g. logical-replication.ts).

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

2148 of 2292 branches covered (93.72%)

Branch coverage included in aggregate %.

140 of 140 new or added lines in 10 files covered. (100.0%)

10489 of 10599 relevant lines covered (98.96%)

136.64 hits per line