Circular paren evaluation

[rebolbot]

Submitted by: Ladislav

The test below crashes the interpreter now, while it did not before.

[
    fact: to paren! []
    insert/only fact fact
    error? try [do fact]
]

^{CC - Data [ Version: alpha 107 Type: Bug Platform: All Category: Unspecified Reproduce: Always Fixed-in:none ]}

[rebolbot]

Submitted by: meijeru

Does not crash in my Windows version of A107 core.

>> fact: to paren! [] insert/only fact fact error? try [do fact]
== true

[rebolbot]

Submitted by: Carl

Agree with Meijeru, does not crash A108.

[rebolbot]

Submitted by: abolka

Does indeed crash with Win32 A107 binaries custom-built from the (unmodified) Host Kit. Does not crash with the pre-built Win32 binaries from RT. Does not crash on Linux for either.

- Host Kit Win32 A107 "Core": crash
- Host Kit Win32 A107 "View": crash

- RT Win32 A107 "Core": no crash
- RT Linux A107 "Core": no crash
- Host Kit Linux A107 "Core": no crash

[rebolbot]

Submitted by: abolka

AdrianS on AltME also confirmed independently:

- Host Kit Win32 (MS compiler) A107 "Core": crash

[rebolbot]

Submitted by: abolka

See also #877 and #1672, which are errors resulting from the same cause (too small a C runtime stack in custom-compiled R3 binaries). #1672 has comments on how to fix/work around this.

[rebolbot]

Submitted by: Ladislav

in the core-tests suite

[rebolbot]

Submitted by: abolka

Works ok on Win32 (and still works with Linux, OSX) with the fix for #1672.