hare

The Hare programming language
git clone https://git.torresjrjr.com/hare.git
Log | Files | Refs | README | LICENSE

commit 5bff8e0a193d4720b375a2c9fc2028500af5c0da
parent 37173d180108300d4c8768322fd03bfcfde70a17
Author: Drew DeVault <sir@cmpwn.com>
Date:   Sat, 26 Jun 2021 13:18:06 -0400

bufio: rewrite README

Signed-off-by: Drew DeVault <sir@cmpwn.com>

Diffstat:
Mbufio/README | 36+++++++++++++++++++++++++++---------
1 file changed, 27 insertions(+), 9 deletions(-)

diff --git a/bufio/README b/bufio/README @@ -1,12 +1,30 @@ bufio provides [[io::stream]] implementations which provide buffered I/O -operations. The [[buffered]] stream will batch reads and writes for an -underlying stream, using a caller-provided buffer as temporary storage. The -[[dynamic]] and [[fixed]] streams produce an [[io::stream]] which reads from or -writes to a data buffer (dynamically and statically allocated, respectively). +support, as well as scanner utility functions which pair well with buffered +streams for maximum efficiency. -Utilities for "scanning" a stream are also provided via [[scantok]] et al. These -will work with any [[io::stream]], but are only efficient on a [[buffered]] -stream. +Two streams are provided which can read from or write to byte slices. [[fixed]] +uses a caller-supplied statically-allocated buffer for storage, producing an +[[io::stream]] which reads from or writes to this buffer. In effect, this allows +the caller to statically allocate a byte array, then produce an [[io::stream]] +which writes to or reads from it. -The use of a [[buffered]] stream also allows the caller to utilize the -[[unread]] family of functions. +The [[dynamic]] stream is similar, but it uses a bufio-managed dynamically +allocated buffer. This creates an [[io::stream]] which efficiently soaks up +writes into a dynamically allocated byte slice, which may be accessed via +[[buffer]] or [[finish]]. The user may also call [[reset]], which empties the +buffer but does not free the underlying storage, allowing the user to re-use the +same buffer for many operations. + +A third stream implementation, [[buffered]], is used to batch read and write +operations against an underlying stream. The caller may use small, frequent read +and write operations, which bufio will batch into larger, less frequent reads +and writes. The caller must supply either one or two temporary buffers for +reading and/or writing, which bufio will use to store future reads, or pending +writes, as necessary. This improves performance when many small reads or writes +would be inefficient, such as when I/O operations require syscalls or network +transmissions. Buffered streams also support an "[[unread]]" operation, which +allows you to "look-ahead" at future data without consuming it from the stream. + +Finally, bufio provides several utilities for "scanning" streams, namely +[[scantok]] et al, which require small, frequent reads, or take advantage of +look-ahead, and thus are most efficient when paired with a [[buffered]] stream.