[emmalloc] Fix verbose output

[kleisauke]

toString() is undeclared.

Fixes a regression introduced in commit 04a0cad.

[kleisauke]

Details:

building:ERROR: Closure compiler run failed:

building:ERROR: /tmp/emscripten_temp_50snxhja/vips-es6.jso3.js:757:85: ERROR - [JSC_UNDEFINED_VARIABLE] variable toString is undeclared
  757|     out("aligned_realloc(ptr=" + ptrToString($0) + ", alignment=" + $1 + ", size=" + toString(Number($2)));
                                                                                            ^^^^^^^^

1 error(s), 0 warning(s)

[sbc100]

Are we missing a test for -sMALLOC=emmalloc-verbose?

[kripken]

Yes, looks like that isn't covered by any existing test. Might be worth adding, but it is a setting for internal debugging.

[kleisauke]

It looks like this is being covered here:

emscripten/test/test_core.py

Line 797 in a6ecc86

'memvalidate_verbose': ['-DEMMALLOC_MEMVALIDATE', '-DEMMALLOC_VERBOSE', '-DRANDOM_ITERS=130'],

emscripten/test/test_browser.py

Line 5229 in a6ecc86

'memvalidate_verbose': (['-sMALLOC=emmalloc-memvalidate-verbose'],),

However, these tests do not verify the debug prints. I'm not sure if that would be useful.

[sbc100]

It looks like this is being covered here:

emscripten/test/test_core.py

Line 797 in a6ecc86

'memvalidate_verbose': ['-DEMMALLOC_MEMVALIDATE', '-DEMMALLOC_VERBOSE', '-DRANDOM_ITERS=130'],

emscripten/test/test_browser.py

Line 5229 in a6ecc86

'memvalidate_verbose': (['-sMALLOC=emmalloc-memvalidate-verbose'],),

However, these tests do not verify the debug prints. I'm not sure if that would be useful.

But you could add --closure=1 which would cause the current code to fail right?

[kleisauke]

But you could add --closure=1 which would cause the current code to fail right?

Ah, good point. Addressed with commit 6fef505, PTAL.

[sbc100]

Oh nice to see this skip being removed!

[sbc100]

I don't think this line should be needed.

Can we instead add EM_JS_DEPS(deps, "$ptrToString"); to emmalloc.c?

[kleisauke]

8118293

[kripken]

lgtm and that test failure on CI looks unrelated.