chore(deps): update dependency fantomas to v8 by renovate[bot] · Pull Request #40 · Shadowsocks-NET/Aurum

renovate · 2026-09-15T18:31:58Z

This PR contains the following updates:

Package	Change	Age	Confidence
fantomas (source)	`6.2.3` → `8.1.1`

Release Notes

fsprojects/fantomas (fantomas)

`v8.1.1`

Compare Source

Fixed

Comment between the type name and the primary constructor is removed. #3569

`v8.1.0`

Compare Source

Changed

Multiline tuple type after a cast or type annotation operator goes one level in. #3561
Line comment lined up with the first element of its container stays with the elements. #3561

Fixed

Parenthesized function type gets an extra indent level where no outer function type needs it. #3554
Multiline tuple in the middle of a function type gives invalid code. #3556
Parenthesized function type after an arrow gives invalid code with indent_size 2. #3557
Long type test with :? gives invalid code. #3558
Line comment after the opening parenthesis of a type moves and breaks the type. #3559
Comment between a match pattern and its when guard gives invalid code. #3560
Regression: Stroustrup bracket style in anonymous record type parameter is not respected. #3069
Space inserted before final > in type parameter. #2860

`v8.0.10`

Compare Source

Fixed

Comment after a property accessor is lost. #3527

`v8.0.9`

Compare Source

Fixed

Comment before a let in the right-hand side of an infix operator is lost. #3522

`v8.0.8`

Compare Source

Changed

Update FCS to 'Add range of the when keyword to SynMatchClauseTrivia', commit 409c7a6 #3531

Fixed

Comment after 'when' on its own line is lost. The syntax tree had no range for the when keyword of a match clause, so a comment after it had no node on its line to attach to. The parser now records that range, and the comment stays after the keyword. #3521

`v8.0.7`

Compare Source

Fixed

Exception abbreviation loses its right-hand side. #3511
Static optimization when ^T struct loses struct. #3514
Static optimization loses the and between its conditions, and reverses its conditions and when clauses. #3516
static extern inside a type loses static. #3515

`v8.0.6`

Compare Source

Changed

Update FCS to 'Record the range of the of keyword in SynUnionCaseTrivia', commit 5e9647f #3502

Fixed

Wrong change of comment's position. The syntax tree had no range for the of keyword of a union case or exception, so a block comment after it, as in exception E of (*name*) string, was attached to the case name and printed in front of of. The parser now records that range, and the comment stays after the keyword. #1959
Cursor is lost when placed on if, then or else if keywords. The cursor now keeps its place on if, elif, else if, then, match, match! and with. #3387
Cursor is lost when placed on the closing bracket of a multiline list or array in the cramped style. #3506

`v8.0.5`

Compare Source

Changed

Improved formatting performance. Formatting the source files of dotnet/fsharp allocates around 40% less memory and is roughly 1.4 times faster than 8.0.4, depending on hardware. The fields of WriterEvent, returned by CodeFormatter.GetWriterEventsAsync, are renamed so that cases share them: text for the strings, amount for IndentBy and UnIndentBy, column for the indent and column cases, and nodeType and range for NodeEnd. #3499
Update FCS to 'Record the range of the exceptionin SynExceptionDefnRepr', commit 349d6aa #3500

Fixed

Comment between doc comment and exception keyword gets shifted. The syntax tree had no range for the exception keyword, so a comment between the XML doc or attributes and the keyword was placed in front of the exception name, after the keyword. The parser now records that range, and the comment stays in front of the keyword. #3483

`v8.0.4`

Compare Source

Changed

Improved formatting performance. Formatting the source files of dotnet/fsharp allocates around 40% less memory and is roughly 1.4 times faster than 8.0.4, depending on hardware. The fields of WriterEvent, returned by CodeFormatter.GetWriterEventsAsync, are renamed so that cases share them: text for the strings, amount for IndentBy and UnIndentBy, column for the indent and column cases, and nodeType and range for NodeEnd. #3499
Update FCS to 'Record the range of the exceptionin SynExceptionDefnRepr', commit 349d6aa #3500

Fixed

Comment between doc comment and exception keyword gets shifted. The syntax tree had no range for the exception keyword, so a comment between the XML doc or attributes and the keyword was placed in front of the exception name, after the keyword. The parser now records that range, and the comment stays in front of the keyword. #3483

`v8.0.3`

Compare Source

Fixed

Second block comment on a line is lost and the next expression moves onto the comment line. A line that holds several comments and no code, such as (* b *) (* c *), was read as one block comment next to code per comment, and there was no code on that line to attach them to. The first one ended up in front of the next expression, pulling it onto the comment line, and the others were dropped. Such a line is now one comment on a line of its own, printed as it was written. #3487
Idempotency problem when a comment follows a multiline infix expression in a let binding. A comment on its own line after c + and d, at the column of both, was attached to d rather than to c + d. That made the expression multiline, so the operator moved to the start of the next line, and on the next format the comment went to the whole expression and it was joined onto one line again. An indented comment now goes after the outermost preceding node that starts a line at its column, and only falls back to a node further along a line when there is none. #3488

`v8.0.2`

Compare Source

Fixed

Code comment inside match-expression leads to additional blank line. A comment that ends the multi-line body of a match clause, and then any blank lines the author kept after it, came out with one blank line more than the source had. The same happened before else, finally, end and the next field of a record, whenever the indented body before them ended in a comment. Since 8.0.0 the unindent that closes such a body leaves the comment's newline at the very end of what was written, and the separator the enclosing construct writes next then opened a second line. Those separators now skip their newline when the comment already wrote one, as the closing brace of a record has done since 8.0.0. Blank lines the author kept, and the ones Fantomas puts between multi-line members and bindings, are unchanged. #3484

`v8.0.1`

Compare Source

Fixed

Comment after a property accessor is lost. #3527

`v8.0.0`

Fixed

Code comment inside match-expression leads to additional blank line. A comment that ends the multi-line body of a match clause, and then any blank lines the author kept after it, came out with one blank line more than the source had. The same happened before else, finally, end and the next field of a record, whenever the indented body before them ended in a comment. Since 8.0.0 the unindent that closes such a body leaves the comment's newline at the very end of what was written, and the separator the enclosing construct writes next then opened a second line. Those separators now skip their newline when the comment already wrote one, as the closing brace of a record has done since 8.0.0. Blank lines the author kept, and the ones Fantomas puts between multi-line members and bindings, are unchanged. #3484

`v7.0.6`

Fixed

Formatting multiple files at once could place comments against the wrong node. #3391

`v7.0.5`

Fixed

Fixed automated GitHub releases. #3205

`v7.0.4`

Changed

Add trusted publishing. #3204

`v7.0.3`

Compare Source

Changed

Improved formatting performance. Measured against 7.0.3, formatting a 2937 line file is roughly 1.5 times faster and allocates around 38% less memory, depending on hardware; part of that speedup is the move to net10.0 rather than this change. The larger figures first reported here, 1.5 to 2 times faster and around 65% less memory, were measured against main at the time, which had drifted well past 7.0.3, so they describe what this change did to the branch rather than what v8 does for someone upgrading. #3388
ServerGarbageCollection is now set on the command line tool's own project file. It was already reaching the tool through Directory.Build.props, which has set it for every project in the repository since v6, so nothing about how the tool runs changed here. An earlier version of this entry claimed Server GC was new in v8; it is not, and v7 shipped with it too. #3388