terser - npm Package Versions

v4.6.2

• A bug where functions were inlined into other functions with scope conflicts has been fixed.
• /*#__NOINLINE__*/ annotation fixed for more use cases where inlining happens.

v4.6.1

• Fixed an issue where a class is duplicated by reduce_vars when there's a recursive reference to the class.

v4.6.0

• Fixed issues with recursive class references.
• BigInt evaluation has been prevented, stopping Terser from evaluating BigInts like it would do regular numbers.
• Class property support has been added

v4.5.1

(hotfix release)

• Fixed issue where () => ({})[something] was not parenthesised correctly.

v4.5.0

• Inlining has been improved
• An issue where keep_fnames combined with functions declared through variables was causing name shadowing has been fixed
• You can now set the ES version through their year
• The output option keep_numbers has been added, which prevents Terser from turning 1000 into 1e3 and such
• Internal small optimisations and refactors

v4.4.3

• Number and BigInt parsing has been fixed
• /*#__INLINE__*/ annotation fixed for arrow functions with non-block bodies.
• Functional tests have been added, using this repository.
• A memory leak, where the entire AST lives on after compression, has been plugged.

v4.4.2

• Fixed a problem with inlining identity functions

v4.4.1

note: This introduced a feature, therefore it should have been a minor release.

• Fixed a crash when unsafe was enabled.
• An issue has been fixed where let statements might be collapsed out of their scope.
• Some error messages have been improved by adding quotes around variable names.

v4.4.0

• Added /*#__INLINE__*/ and /*#__NOINLINE__*/ annotations for calls. If a call has one of these, it either forces or forbids inlining.

v4.3.11

• Fixed a problem where window was considered safe to access, even though there are situations where it isn't (Node.js, workers...)
• Fixed an error where ++ and -- were considered side-effect free
• Number(x) now needs both unsafe and and unsafe_math to be compressed into +x because x might be a BigInt
• keep_fnames now correctly supports regexes when the function is in a variable declaration