with-hash-table-iterator CI fail

[kchanqvq]

Hmm, weird, it passes tests fine locally.

Also once this is ready we could implement maphash on top of this, but do we still want to keep forEach for potential JS users?

[davazp]

It failing for me locally too

perform test/hash-tables.lisp!
Test `(LET ((JSCL::H (MAKE-HASH-TABLE))) (SETF (GETHASH (QUOTE JSCL::FOO) JSCL::H) 1) (SETF (GETHASH 2 JSCL::H) (QUOTE JSCL::BAR)) (WITH-HASH-TABLE-ITERATOR (JSCL::I JSCL::H) (AND (JSCL::MV-EQL (JSCL::I) T JSCL::FOO 1) (JSCL::MV-EQL (JSCL::I) T 2 JSCL::BAR) (JSCL::MV-EQL (JSCL::I) NIL NIL NIL))))' failed.

Test `(LET ((JSCL::H (MAKE-HASH-TABLE :TEST (QUOTE EQUAL)))) (SETF (GETHASH "foo" JSCL::H) "test") (SETF (GETHASH "bar" JSCL::H) 2) (WITH-HASH-TABLE-ITERATOR (JSCL::I JSCL::H) (AND (JSCL::MV-EQL (JSCL::I) T "foo" "test") (JSCL::MV-EQL (JSCL::I) T "bar" 2) (JSCL::MV-EQL (JSCL::I) NIL NIL NIL))))' failed.

perform test/iter-macros.lisp!
perform test/list.lisp!
perform test/macros.lisp!
perform test/misc.lisp!
perform test/multiple-values.lisp!
perform test/numbers.lisp!
perform test/package.lisp!
perform test/print.lisp!
perform test/read.lisp!
perform test/seq.lisp!
perform test/setf.lisp!
perform test/stream.lisp!
perform test/strings.lisp!
perform test/types.lisp!
perform test/variables.lisp!
perform test/loop/validate-tests.lisp!
perform test/loop/base-tests.lisp!


Finished. The execution took 0.23 seconds.

1704/1706 test(s) passed successfully.

[davazp]

@kchanqvq maybe you didn't rerun build-all or ./make.sh before run-tests.sh?

I usually do ./make.sh && ./run-tests.sh before I commit something.

[davazp]

@kchanqvq Found the problem. I (and the pipeline) was using a too old version of node. Feel free to upgrade it in the pipeline.

According to this we need node 22 at least.

I don't mind requiring it. But we should probably document the requirement somewhere. Someone might be using jscl for some app that supports older browsers or node.

[kchanqvq]

@kchanqvq Found the problem. I (and the pipeline) was using a too old version of node. Feel free to upgrade it in the pipeline.

According to this we need node 22 at least.

I don't mind requiring it. But we should probably document the requirement somewhere. Someone might be using jscl for some app that supports older browsers or node.

Ah, I should use some older feature, then. Symbol.iterator is almost universally supported, but I didn't use it because our xstring and consequently oget etc destroys Symbols. Maybe I should change xstring so that it pass through Symbols? But then the name becomes a bit confusing.

[kchanqvq]

Maybe I should change xstring so that it pass through Symbols? But then the name becomes a bit confusing.

@davazp opinions?

[davazp]

You basically want (oget! x symbol) to work, right?

Can we wait until we see what we do with the FFI? We could add a comment or check in case Iteartor:from is not available for now and upgrade the pipeline to 22.

I plan to have the basics of the FFI ready soon. Then we could change it so we also support node 20.

[kchanqvq]

Can we wait until we see what we do with the FFI?

Of course.

[davazp]

I'll update the pipeline to 22 for now.