IMO it’s reasonable to warn. It’s more likely to be an error on the programmer’s part than a legitimate need. If the programmer really wants to match against quote, they can write it out (and IMO they should).

Agreed

yes, given that 'a and (quote a) are the same, it makes sense to direct them to the expanded version

OK, the issue is implemented in clj-kondo: https://twitter.com/borkdude/status/1471936924483006475 by @U02QXKSHP97 (thanks!) I just wonder if the warning could become more clear to explain what the issue is. Currently users might still think: yeah, quoting, that's what I want dude, what are you complaining about.

Maybe we should emphasize: > The test-constants are not evaluated. They must be compile-time > literals, and need not be quoted.

So the message could become: "Test-constants are not evaluated and need not be quoted: 'a"

should the condp fragment quote the set in (case '#{a} #{a} 1 :foo 2 'a 3) ?

this example is super confusing... I'm still trying to mentally figure it out 🙂

user=> (#'clojure.core/shift-mask 9 3 (.hashCode '#{a}))
1
user=> (#'clojure.core/shift-mask 9 3 (.hashCode 'quote))
3
user=> (#'clojure.core/shift-mask 9 3 (.hashCode 'a))
1
user=> (#'clojure.core/shift-mask 9 3 (.hashCode :foo))
2

those are the 4 branches in the example, note two clashes

right

but it seems the condp expansion isn't quoting the set, which causes an error?

note that this is a totally fabricated example, not one from production code, but still

The 'a is a bit confusing, because in any other context it would be a single value, but in a case it becomes a pair. But I agree with Michiel, the condp isn’t quoting the values that it receives. It’s not just a missing quote on that set. It also affects the symbol a:

user=> (case 'a #{a} 1 :foo 2 a 3)
Syntax error compiling at (REPL:1:1).
Unable to resolve symbol: a in this context
user=> (def a nil)
#'user/a
user=> (case 'a #{a} 1 :foo 2 a 3)
Execution error (IllegalArgumentException) at user/eval158 (REPL:1).
No matching clause: a

I’ve put a suggested patch on https://ask.clojure.org/index.php/11395/hash-collisions-case-statements-try-evaluate-the-constants, since I think it has to go through there before it can go to Jira, right?

Thanks!

^"[Ljava.lang.String;"

Ljava.lang.String;

dorabs-imac:~ dorab$ clj
Clojure 1.10.3
user=> (type (into-array String []))
[Ljava.lang.String;
user=> 

https://docs.oracle.com/javase/specs/jvms/se7/html/jvms-4.html#jvms-4.3.2-200

You need the quotes because of metadata rules, you need the [ because it’s an array, and you need the ; because it’s a reference type

that's some wild syntax, thanks!

Unrelated to @U0ELT5ZDE’s question, this is an interesting rabbithole to go down. That link above is for ClassNames in internal form for the JVM. That’s actually different to the form of BinaryNames described in the Java Language Specification. Quoting from the JVM spec: > For historical reasons, the syntax of binary names that appear in `class` file structures differs from the syntax of binary names documented in JLS §13.1. In this internal form, the ASCII periods (`.`) that normally separate the identifiers which make up the binary name are replaced by ASCII forward slashes (`/`). The identifiers themselves must be unqualified names (https://docs.oracle.com/javase/specs/jvms/se7/html/jvms-4.html#jvms-4.2.2). The Java Language Specification https://docs.oracle.com/javase/specs/jls/se17/html/jls-13.html#jls-13.1 is a bit harder to read than the JVM spec document, so it makes sense to see what they look like in the JVM spec, and just replace the / characters with . characters.

You can also get to these links by looking at the javadoc for https://docs.oracle.com/en/java/javase/17/docs/api/java.base/java/lang/Class.html#getName(). This says that it will return the binary name of the class with a https://docs.oracle.com/en/java/javase/17/docs/api/java.base/java/lang/ClassLoader.html#binary-name. This describes class names with dot separators, and has links to the Java Language Specification (https://docs.oracle.com/javase/specs/jls/se17/html/jls-6.html#jls-6.7 and https://docs.oracle.com/javase/specs/jls/se17/html/jls-13.html#jls-13.1)

@U07MMB5JB close - you need the double quotes because otherwise it wouldn't be readable (unbalanced [ token and trailing ; token break reader rules)