Compiler hangs when compiling large string concatenation

[zeitbit]

Nim Compiler Version 2.3.1 [Windows: amd64]
Compiled at 2025-06-09.

const CRLF = "\r\n"

proc Test(): string =
    let text = "---------------------" &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF &
    "----------" & "----------" & CRLF

    text

echo Test()

Description

Current Output

Expected Output

Known Workarounds

No response

Additional Information

No response

[metagn]

Caused by quadratic semchecks of the operands allocating too much memory with copyTree to get an nOrig. Semchecking a call semchecks each of its arguments individually, and when the arguments are calls themselves, the arguments of those are semchecked as well. This keeps calling copyTree.

The orig stuff seems to only be used for semantic code for the lifting of do: blocks to lambdas which is deprecated, going to test removing it, or at least making a shallow copy.

Edit: Never mind, the quadratic recursions still cause other issues even without the copying. The copying is just the first result.

[forchid]

@zeitbit It always passes in nim 1.6.20 with --mm:refc, --mm:orc or --mm:arc. This issue also existing in nim 2.2.4.

nim 1.6.20 test case

basics>nim c -d:release largestrcon
Hint: used config file '...\.choosenim\toolchains\nim-1.6.20\config\nim.cfg' [Conf]
Hint: used config file '...\.choosenim\toolchains\nim-1.6.20\config\config.nims' [Conf]
..........................................................
CC: .../.choosenim/toolchains/nim-1.6.20/lib/std/private/digitsutils.nim
CC: .../.choosenim/toolchains/nim-1.6.20/lib/system/dollars.nim
CC: .../.choosenim/toolchains/nim-1.6.20/lib/system/io.nim
CC: .../.choosenim/toolchains/nim-1.6.20/lib/system.nim
CC: largestrcon.nim
Hint:  [Link]
Hint: gc: refc; opt: speed; options: -d:release
26830 lines; 4.298s; 48.703MiB peakmem; proj: largestrcon; out: ...\basics\largestrcon.exe [SuccessX]

basics>largestrcon
-----------------------------------------
--------------------
--------------------
--------------------
--------------------
--------------------
--------------------
--------------------
...
...
...
basics>