RFR: 8367530: The exhaustiveness errors could be improved [v13]

[Maurizio Cimadamore]

On Mon, 26 Jan 2026 11:58:07 GMT, Maurizio Cimadamore <mcimadamore at openjdk.org> wrote:

>> src/jdk.compiler/share/classes/com/sun/tools/javac/util/AbstractDiagnosticFormatter.java line 237:
>> 
>>> 235:                                                   StringUtils.toLowerCase(tag.name()));
>>> 236:         }
>>> 237:         else if (arg instanceof BindingPattern bp) {
>> 
>> Have you tried playing with the Formattable interface? This is used by diagnostic formatters as an escape hatch to provide custom formatting -- but probably that's not good enough, as Formattable expects to just map an object into a string w/o the help of a formatter -- which is clearly not the case here. Maybe just having an abstract `String format(AbstractDiagnosticFormatter)` in PatternDescription could be enough -- then the code here doesn't have to worry about all possible cases.
>
> That said, in Rich formatter we also need to preprocess, so perhaps keeping the code as is here would be more symmetric...

I suppose overall I'm not entirely sure if, at this point, pattern descriptions should be pattern descriptions, or already translated into some kind of fragment -- because the code in formatters can deal with fragments recursively very well -- but can do less so with objects.

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/27256#discussion_r2727338487