[WIP] Fix Issue 18721 - ICE in dmd/cond.d(378) when compiling static foreach with -D

[wilzbach]

Okay, so yet another noobish attempt.
The ICE is caused by the invalid template never been instantiated and run through semantic.
DDoc, however, tries to inspect the template with getEponymousMember and thus obviously spectactuarly fails.

Has anyone a better idea to tackle this?

[dlang-bot]

Thanks for your pull request, @wilzbach!

Bugzilla references

Auto-close	Bugzilla	Severity	Description
✓	18721	critical	ICE in dmd/cond.d(378) when compiling static foreach with -D

⚠️⚠️⚠️ Warnings ⚠️⚠️⚠️

• Regression or critical bug fixes should always target the stable branch. Learn more about rebasing to stable or the D release process.

To target stable perform these two steps:

1. Rebase your branch to upstream/stable:

git rebase --onto upstream/stable upstream/master

2. Change the base branch of your PR to stable

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#8128"

[jacob-carlborg]

Is static foreach run before template instantiation?

[stefan-koch-sociomantic]

static foreach is supposed to be run as soon as possible.

@wilzbach static foreach can't be lowered`. I have no idea what you mean with that.

[wilzbach]

Is static foreach run before template instantiation?

AFAICT yes and that's the problem.:

core.exception.AssertError@dmd/cond.d(378): Assertion failure
----------------
??:? _d_assertp [0x74903d]
dmd/cond.d:378 pure nothrow @nogc @safe void dmd.cond.StaticForeach.prepare(dmd.dscope.Scope*).__require() [0x51392f]
dmd/cond.d:377 void dmd.cond.StaticForeach.prepare(dmd.dscope.Scope*) [0x5137dc]
dmd/attrib.d:1083 _ZN24StaticForeachDeclaration7includeEP5Scope [0x50c2df] <--- so we ended in an invalid static foreach block
dmd/doc.d:1042 _ZN11emitComment11EmitComment5visitEP17AttribDeclaration [0x55811b] <--- finds static foreach here and tries to resolve it

(the rest of the stack trace)

dmd/parsetimevisitor.d:75 _ZN16ParseTimeVisitorI10ASTCodegenE5visitEP24StaticForeachDeclaration [0x6437f2]
dmd/attrib.d:1143 _ZN24StaticForeachDeclaration6acceptEP7Visitor [0x50c48d]
dmd/doc.d:1091 void dmd.doc.emitComment(dmd.dsymbol.Dsymbol, dmd.root.outbuffer.OutBuffer*, dmd.dscope.Scope*) [0x557798]
dmd/doc.d:796 void dmd.doc.emitMemberComments(dmd.dsymbol.ScopeDsymbol, dmd.root.outbuffer.OutBuffer*, dmd.dscope.Scope*) [0x55767c]
dmd/doc.d:916 _ZN11emitComment11EmitComment4emitEP5ScopeP7DsymbolPKc [0x557c57]
dmd/doc.d:960 _ZN11emitComment11EmitComment5visitEP11Declaration [0x557ded]
dmd/parsetimevisitor.d:47 _ZN16ParseTimeVisitorI10ASTCodegenE5visitEP16AliasDeclaration [0x643492]
dmd/declaration.d:929 _ZN16AliasDeclaration6acceptEP7Visitor [0x535d61]
dmd/doc.d:1091 void dmd.doc.emitComment(dmd.dsymbol.Dsymbol, dmd.root.outbuffer.OutBuffer*, dmd.dscope.Scope*) [0x557798]
dmd/doc.d:796 void dmd.doc.emitMemberComments(dmd.dsymbol.ScopeDsymbol, dmd.root.outbuffer.OutBuffer*, dmd.dscope.Scope*) [0x55767c]
dmd/doc.d:439 _Z10gendocfileP6Module [0x556783]
dmd/mars.d:911 int dmd.mars.tryMain(ulong, const(char)**) [0x5f8097]

@wilzbach static foreach can't be lowered`. I have no idea what you mean with that.

Well, I would prefer an error from semantic like:

fail_compilation/fail18721.d(12): Error: undefined identifier T
fail_compilation/fail18721.d(18): Error: template instance `fail18721.allSameType!()` error instantiating

But the way I see this bug is that semantic never runs and DDoc still tries to resolve the invalid static foreach, so I improvised with the error message. I'm more than open to better ideas.

[jacob-carlborg]

static foreach is supposed to be run as soon as possible.

I though it would be run at the same time as static if, which does not have this problem.

[wilzbach]

FTW I accidentally found another Bugzilla issues that segfault on the same line in the Json generator:

https://issues.dlang.org/show_bug.cgi?id=18581

[WalterBright]

The actual error looks like T is undefined.

[wilzbach]

This has been fixed in master already. Closing.