Fix GH-15901: phpdbg: Assertion failure on i funcs

[cmb69]

New hash tables are not automatically packed, so we must not treat them as such. Therefore we guard the foreach appropriately.

[iluuu1994]

HT_IS_INITIALIZED() would make slightly more sense to me, but this looks fine too.

[cmb69]

HT_IS_INITIALIZED() would make slightly more sense to me, […]

Ah, right, when reading the code this is better. However, for maintenance using HT_IS_PACKED() might be better, since using ZEND_HASH_PACKED_FOREACH_PTR on an arbitrary initialized array would not work (I think).

[iluuu1994]

Indeed, but then at least we'd crash, rather than just silently not hand the array.

[cmb69]

Makes sense. Changed to HT_IS_INITIALIZED().

[cmb69]

The macOS CI failure seems to be unrelated to the PR.

[nielsdos]

LGTM
and yeah, curl has started failing on macOS since yesterday or so, it's unrelated

[cmb69]

curl has started failing on macOS since yesterday or so

Should we update the PHP-8.2 macOS CI?

[iluuu1994]

@cmb69 Do you know what the cause is? I have no clue. The failures are seemingly random, although it's always the same tests that fail.

[cmb69]

@iluuu1994, hmm, just guessing, but might be due to https://github.com/php/php-src/pull/15877/files#diff-ef4c5569bff099e37e482441804e39e99ca75e1a0cdb13bf555a99a2f4feef3cL21.

[iluuu1994]

I see around 20 successfully macOS builds after that commit, so it seems unlikely. But who knows?

[cmb69]

See #15958. If that fails, it's something else; if it succeeds, well, who knows? :)

[cmb69]

Well, #15958 failed; we may continue the discussion there.