Allow for missing invisible close delim when reparsing an expression.

[nnethercote]

This can happen when invalid syntax is passed to a declarative macro. We shouldn't be too strict about the token stream position once the parser has rejected the invalid syntax.

Fixes #139248.

r? @petrochenkov

---

try-job: test-various

[petrochenkov]

@bors r+ rollup

[bors]

📌 Commit a2a081a has been approved by petrochenkov

It is now in the queue for this repository.

[nnethercote]

#137874 is similar to #139248, there might be a common fix slightly different to what I've done here.

@bors r-

[nnethercote]

#137874 is similar to #139248, there might be a common fix slightly different to what I've done here.

No, it needs a separate fix.

@bors r=petrochenkov

[bors]

📌 Commit a2a081a has been approved by petrochenkov

It is now in the queue for this repository.

[Zalathar]

Possibly failed in rollup: #139344 (comment)

@bors r-

[Zalathar]

@bors try

[bors]

⌛ Trying commit a2a081a with merge 46cc200...

[bors]

💔 Test failed - checks-actions

[nnethercote]

I don't understand what's going wrong. The test passes locally and the diff on CI is strange:

41    |
42 LL | thread_local! { static a : () = (if b) }
43    | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ expected expression
-   --> $SRC_DIR/std/src/sys/thread_local/native/mod.rs:LL:COL
45    |
46    = note: while parsing argument for this `expr` macro fragment
47    |

[Zalathar]

Oh, I wonder if this is wasm causing problems because the path of the source file defining the macro is different.

In the raw stderr, the path on line 44 is:

    --> /rustc/FAKE_PREFIX/library/std/src/sys/thread_local/statik.rs:28:18

[nnethercote]

How is wasm involved?

I could rewrite the test to use a local macro instead of thread_local!, that might help.

[nnethercote]

@bors try

[bors]

⌛ Trying commit ad9d8f9 with merge 281c69c...

[nnethercote]

@bors try

[bors]

⌛ Trying commit 296b484 with merge 9840dcf...

[nnethercote]

@bors try

[bors]

⌛ Trying commit fbdf9f8 with merge 732ba88...

[bors]

☀️ Try build successful - checks-actions
Build commit: 732ba88 (732ba88caf76736237d422d1580a6865c071b6ff)

[nnethercote]

I could rewrite the test to use a local macro instead of thread_local!, that might help.

That worked. I still don't understand what went wrong with the thread_local! version.

@bors r=petrochenkov

[bors]

📌 Commit fbdf9f8 has been approved by petrochenkov

It is now in the queue for this repository.

[Zalathar]

That worked. I still don't understand what went wrong with the thread_local! version.

I believe the problem was:

• When using thread_local!, the error message includes a path into standard library source where thread_local_inner is defined ($SRC_DIR/std/src/sys/thread_local/native/mod.rs:LL:COL).
• Some targets use a different definition of thread_local_inner that is in a different file, so the path into standard library source is different.
• That different path causes the snapshots to not match, even though the difference isn't important to what this test is testing.

All of that is a distraction from what this test actually cares about, so I think switching to a local macro was the right call.

[nnethercote]

so I think switching to a local macro was the right call.

Thanks for the explanation!

[petrochenkov]

This is included into #139464 so I'll close.