Version 3.0.1

This is a bug-fix release.

Changelog

• Fixed another bug in get_num_codepoints (#16)
• Fixed bug in make.bat so that it works now 😃
• Fixed bug in comparison functions (#23)
• Renamed namespace detail to tiny_utf8_detail in order to avoid accidental namespace clashes

Version 3

This release does not change any behaviour. However, from this release on, tiny-utf8 is header-only.

Version 2.2

This is a bug-fix as well as an enhancement release including API changes.

Changelog

• Marked utf8_string( value_type ) as explicit (thanks to #21).
• Changed return type of utf8_string::compare from difference_type to int and fixed #20.
• Broadened support for char/value_type literals with embedded zeros.
• Added comparison functions and operators for literals and std::string.
• Removed utf8_string::not_equals.
• Fixed bug in utf8_string::get_codepoint_bytes.
• Evaluated execution statistics and adjusted the LUT-creation biases from 50+/-25% to 25+-8%.
• Added public function utf8_string::lut_active() for curiosity about whether the Look-Up-Table is currently active.

Version 2.1.1

This is a bug-fix as well as a minor enhancement release.

Changelog

• Fixed bug in get_num_codepoints (#16)
• Added CMake support

Version 2.1

This is a bug-fix release as well as a minor feature enrichment.

Changelog

• Whether or not an utf8_string instance keeps, creates or drops a Look-Up-Table (LUT) now takes into account whether a LUT is already present. Threfore, if an instance has a LUT, the #multibyte/#codepoint ratio needs to exceed 62,5% in order for the instance to drop the LUT. Vice-versa, if an instance lacks a LUT, the ratio needs to drop below 37,5% in order for the instance to start a LUT.
• Fixed bug in the constructor taking a value_type with codepoints > 127 (#15)
• Fixed bug in raw_insert, raw_replace and append, when new_buffer_size is within a certain integer domain and new_buffer_size*2 is not (#14)

Version 2.0.3

This is a bug-fix release.

Changelog

• Fixed bug in utf8_string::get_num_codepoints(), if SSO (Small-String-Optimization) is active (#12)
• Fixed bug in the default constructor, where the string data was not initialized with a leading zero (#10)
• Fixed compilation issues with VS 2015 (#11, #9)

Version 2.0.2

This is a minor release, not changing any functionality.

Changelog

• Made public the following methods:
  • get_num_bytes
  • get_num_bytes_from_start
  • get_num_codepoints

Version 2.0.1

This is a minor bug-fix release.

Changelog

• Fixed #5 (that is, utf8_string::swap())

Version 2.0

This is a major release. Some functions have been renamed, the memory layout has been updated and many bugs have been fixed.

Changelog

• Fixed many many bugs.
• Added .assign() and std::initializer_list-ctor.
• Made iostream-support the default.
• Dropped ANSI-Support in favor of execution speed.
• Dropped the is_malformed() function.
• Dropped the num_multibytes() function.
• Added operator=() that can reuse old memory.
• Applied some intense optimizations!

Version 1.5

This is a bug-fix release.

Changelog

• utf8_string::replace was fixed in very many spots