udsv

                                                               M. Tuddenham
                                                               August 2023

                               UDSV SerDe

1. Introduction

This is a rust serde library for the UNIX Delimiter Separated Values (UDSV) file format.

2. Records vs File

record_to_str, and record_from_str handle just one record. In the future there will be to_str and from_str to handle a whole file of records.

3. Non-specified data types

While Numbers/Enums/Structs/Tuples aren't in the original specification, an attempt has been made to best (de)serialise them here.

3.1 Numbers

To do.

3.2 Enums Enums are always externally tagged.

3.3 Struct Structs are flattened, and the fields are colon-separated. Structs are never tagged.

3.4 Tuples

Tuples are (de)serialised identically to lists.

3.5 Options

None is serialised as the empty string, Some(x) is just x serialised. Options are lossy, i.e. there is no distinction between Some(()), Some(""), and None.

3.6 Booleans

Booleans are serialised as the strings "true" and "false".

4. To do list

• Allow (de)serialising many records e.g. from a file.
• Add tracing to tests https://docs.rs/tracing-test/latest/tracing_test/
• Better errors - provide additional information in its error type, for example the line and column at which the error occurred, the byte offset into the input, or the current key being processed.
• Ser
  • Disallow nesting
• De

  • Allow derseriliasing into a &str - https://serde.rs/lifetimes.html

  • Disallow nesting

  • floats

  • signed ints

  • int/float overflow