Warning: Work in progress! Leave feedback on Zulip or Github if you'd like this doc to be updated.

Solutions to the Framing Problem

How do you write multiple records to a pipe, and how do you read them?

You need a way of delimiting them. Let's call this the "framing problem" — a term borrowed from network engineering.

This doc categorizes different formats, and shows how you handle them in YSH.

YSH is meant for writing correct shell programs.

Table of Contents
A Length Prefix
Solutions Using a Delimiter
Fixed Delimiter: Newline or NUL byte
Chosen Delimiter: Here docs and multipart MIME
C-Style \ escaping allows arbitrary bytes
Escaping-Based Records
Conclusion

A Length Prefix

Netstrings are a simple format defined by Daniel J Bernstein.

3:foo,  # ASCII length, colon, byte string, comma

This format is easy to implement, and efficient to read and write.

But the encoded output may contain binary data, which isn't readable by a human using a terminal (or GUI). This is significant!


TODO: Implement read --netstr and write --netstr

Solutions Using a Delimiter

Now let's look at traditional Unix solutions, and their problems.

Fixed Delimiter: Newline or NUL byte

In traditional Unix, newlines delimit "records". Here's how you read them in shell:

while IFS='' read -r; do  # confusing idiom!
  echo line=$REPLY
  break                   # remaining bytes are still in the pipe
done

YSH has a simpler idiom:

while read --raw-line {   # unbuffered
  echo line=$_reply
  break                   # remaining bytes are still in the pipe
}

Or you can read all lines:

for line in (io.stdin) {     # buffered
  echo line=$line
  break                   # remaining bytes may be lost in a buffer
}

However, in Unix, all of these strings may have newlines:


But these C-style strings can't contain the NUL byte, aka \0. So GNU tools have evolved support for another format:

find . -print0  # write data
xargs -0        # read data; also --null
grep -z         # read data; also --null-data
sort -z         # read data; also --zero-terminated
                # (Why are all the names different?)

In Oils, we added a -0 flag to read to understands this:

$ find . -print0 | { read -0 x; echo $x; read -0 x; echo $x; }
foo  # could contain newlines!
bar

Chosen Delimiter: Here docs and multipart MIME

Shell has has here docs that look like this:

cat <<EOF
the string EOF
can't start a line
EOF

So you choose the delimiter, with the "word" you write after <<.


Similarly, when your browser POSTs a form, it uses MIME multipart message format:

MIME-Version: 1.0
Content-Type: multipart/mixed; boundary=frontier

This is a message with multiple parts in MIME format.
--frontier
Content-Type: text/plain

This is the body of the message.
--frontier

So again, you choose a delimiter with boundary=frontier, and then you must recognize it later in the message.

C-Style \ escaping allows arbitrary bytes

JSON can express strings with newlines:

"line 1 \n line 2"

It can also express the zero code point, which isn't the same as NUL byte:

"zero code point \u0000"

[J8 Notation][] is an extension of JSON that fixes this:

"NUL byte \y00"

(We use \y00 rather than \x00, because Python and JavaScript both confuse \x00 with U+0000. The zero code point may be encoded as 2 or 4 NUL bytes.)

Escaping-Based Records

TSV files are based on delimiters, but they aren't very readable in a terminal.

TODO

So TSV8 offers and "aligned" format:

#.ssv8 flag      desc                 type
type   Str       Str                  Str
       --verbose "do it \t verbosely" bool
       --count   "count only"         int

So this format combines two strategies:

Conclusion

Traditional shells mostly support newline-based records. YSH supports:

  1. Length-prefixed records
  2. Delimiter-based records
  1. Escaping-based records with JSON and the [J8 Notation][] extension.
Generated on Mon, 28 Oct 2024 16:48:32 +0000