account for 'unsigned long' name mangling in OSX 32

[WalterBright]

No description provided.

[dlang-bot]

Thanks for your pull request, @WalterBright!

Bugzilla references

Your PR doesn't reference any Bugzilla issue.

If your PR contains non-trivial changes, please reference a Bugzilla issue or create a manual changelog.

Testing this PR locally

If you don't have a local development environment setup, you can use Digger to test this PR:

dub fetch digger
dub run digger -- build "master + dmd#8243"

[WalterBright]

ping @braddr

[wilzbach]

If this is required for the auto-tester to upgrade to a newer DMD version, targeting stable might help (and tests of course, too).

[WalterBright]

This is needed to compile dmd with 2.079 on OSX. No new tests are needed, because it failed the autotester build the way it is now.

[WalterBright]

Blocking #8112

[JinShil]

I'd be happy to review this, but I need some context. What's the problem and how does this fix it?

[WalterBright]

The problem comes when using dmd 2.079 as the boot compiler on OSX:

https://auto-tester.puremagic.com/show-run.ghtml?projectid=1&runid=3161515&isPull=true

Undefined symbols for architecture x86_64:
  "Identifier::idPool(char const*, unsigned long)", referenced from:
      test_visitors() in cxxfrontend.o
      test_semantic() in cxxfrontend.o
  "IntegerExp::create(Loc, unsigned long, Type*)", referenced from:
      test_visitors() in cxxfrontend.o
ld: symbol(s) not found for architecture x86_64

I have fixes for the unsigned long mangling issue itself, but those aren't in the 2.079 yet. Hence this PR which is a workaround.

[schveiguy]

but those aren't in the 2.079 yet

should we target a later base compiler then?

[WalterBright]

should we target a later base compiler then?

It's so hard getting this moving forward with all the hand-wringing, I mean months and months and months of delays, I do not wish to start over with a new version. Let's just get this done.

[wilzbach]

FWIW if this is already fixed in 2.080, then there's nothing stopping Brad from using that one...
(the only guarantee we agreed to give is that the latest release can compile DMD)

Of course I don't object to this PR.

[WalterBright]

I'll take that as approval :-)