Formatting breaks pattern needed for null-conditonal operators by adding spaces #2066

Activity

1. Dylan Vos (ThaDaVos)

   added

   Issue-Bug

   on Sep 14, 2024
2. andyleejordan commented on Feb 25, 2025

   

   Andy Jordan (andyleejordan)

   on Feb 25, 2025

   Member

   More actions

   Hi there, this looks to me like a potential bug in PSScriptAnalyzer, specifically the part of the UseConsistentWhitespace rule that looks for spaces within braces: https://github.com/PowerShell/PSScriptAnalyzer/blob/main/docs/Rules/UseConsistentWhitespace.md#checkinnerbrace-bool-default-value-is-true

   I'm going to transfer to that repo, it would be helpful if you can confirm by enabling/disabling that rule.
3. Andy Jordan (andyleejordan)

   transferred this issue fromPowerShell/vscode-powershellon Feb 25, 2025
4. liamjpeters commented on Nov 3, 2025

   

   Liam Peters (liamjpeters)

   on Nov 3, 2025

   Last edited by liamjpeters

   Contributor

   More actions

   It's definitely the CheckInnerBrace part of UseConsistentWhitespace. Turning it on and off for a simple case, $a.{Prop}, confirms it.

   The curly-braced member access (not sure on any official naming) isn't parsed into any special AST node structure, and the token stream is just basically VARIABLE, DOT, LCURLY, IDENTIFIER, RCURLY.

   Accounting for this would be a case of looking for the braces that are used in this way and ignoring them when it comes time to formatting.

   It's made complex by the below being valid:

   $a = [PSCustomObject]@{
       'Prop' = 5
       '{Prop}' = 'Value'
       '$a.{Prop}' = 'Value2-electric-boogaloo'
   }
   $a.{{Prop}}     # prints 'Value'
   $a.{$a.{Prop}}  # prints 'Value2-electric-boogaloo'

   So nothing contained within the curly-braced member access braces should be formatted at all as it breaks it.

   One strategy could be:

   • Keep a list of excluded extents/ranges.
   • Iterate over tokens, for each dot token:
     • Check the dots extent isn't in an excluded extent/range
     • Check the dot has a preceding significant token (ignoring comments and skipping on line continuations and new lines) with no whitespace or new-line gaps. Variables, identifiers, literals, braces - all appear to be acceptable.
     • Check the dot's following significant token (ignoring comments, new lines, line continuations) is a LCURLY.
     • Find the matching RCURLY for this LCURLY.
     • Add an extent/range to the ignored list that goes from the LCURLY to the RCURLY.
   • When considering a brace for a whitespace addition, check first that it's not within an ignored extent/range.

   Andy Jordan (@andyleejordan) - Does that sound like a reasonable approach or am I over-complicating it and missing something simpler?
5. Andy Jordan (andyleejordan)

   closed this as completedin #2140on Dec 3, 2025