v0.3.0

• Added Bash support.

• Added cb.code command that simply displays code and executes nothing.

• Added cb.paste command that allows code and/or output to be copied from
  other code chunks.

• Added markup option for show and hide. This displays the Markdown
  source for inline code or a code block.

• Added support for naming code chunks with the name keyword. Added support
  for copying named code chunks into other code chunks with the copy
  keyword.

• Runtime source errors (code is improperly divided into code chunks, such as
  complete=true when it is not) are now handled like any other source
  errors, rather than as a special instance of stderr. As part of this, the
  errors now have their own entry in the cache.

• FileNotFoundError for subprocesses now returns FailedProcess with
  correct attribute values.

• Fixed compatibility with languages that do not define an inline expression
  formatter.

• Fixed stderr syncing bug for languages that have multiple line number
  patterns.

• Fixed compatibility with Pandoc commands in which output format is inferred
  from output file name. Better output when codebraid is run with no
  arguments (#3).

• Code that interferes with Codebraid's templates is now detected and results
  in error messages.

• Session names are now restricted to identifier-style strings.

• In language definitions, field tempsuffix is renamed to temp_suffix.

• Raw output from code blocks no longer merges with a following paragraph or
  other block.

• Improved newline handling. All text is processed in universal newlines
  mode. Only \n is treated as a newline for line splitting. This avoids
  edge cases from str.splitlines().

• outside_main is now properly checked for compatibility with other options.

• More efficient AST processing.