Chapel: Internal error when returning an array from function whose return type is not an array

[buddha314]

During a test, I got an "it's us, not you" kinda error, like I got from my stupid girlfriend in high school. So I will just say, Chapel, you were lucky to have me! And here is my error

[Removing log file with duplicate name /Users/buddha/github/chapel/test/Logs/buddha.darwin.log
[Starting Chapel regression tests - 180206.140349]
[Warning: start_test not invoked from a subdirectory of $CHPL_HOME]
[starting directory: "/Users/buddha/github/buddha314/numsuch/test"]
[Logs directory: "/Users/buddha/github/chapel/test/Logs"]
[log_file: "/Users/buddha/github/chapel/test/Logs/buddha.darwin.log"]
[CHPL_HOME: /Users/buddha/github/chapel]
[host platform: darwin]
[target platform: darwin]
[valgrind: OFF]
[performance tests: OFF]
[compiler performance tests: OFF]
[number of trials: 1]
[performance graph generation: OFF]
[compiler: "/Users/buddha/github/chapel/bin/darwin/chpl"]
[compopts: "--cc-warnings"]
[execopts: ""]
[launchcmd: ""]
[comm: "none"]
[localeModel: "flat"]
[numlocales: "(default)"]
[max_locales: "(default)"]
[tests: "/Users/buddha/github/buddha314/numsuch/test/NamedMatrix-test.chpl"]
[directories: ""]

### Chapel Environment ###
machine info: Darwin macbuddha.lan 16.7.0 Darwin Kernel Version 16.7.0: Mon Nov 13 21:56:25 PST 2017; root:xnu-3789.72.11~1/RELEASE_X86_64 x86_64
CHPL_HOME: /Users/buddha/github/chapel *
script location: /Users/buddha/github/chapel/util
CHPL_TARGET_PLATFORM: darwin
CHPL_TARGET_COMPILER: clang
CHPL_TARGET_ARCH: native
CHPL_LOCALE_MODEL: flat *
CHPL_COMM: none *
CHPL_TASKS: qthreads
CHPL_LAUNCHER: none *
CHPL_TIMERS: generic
CHPL_UNWIND: none
CHPL_MEM: jemalloc
CHPL_ATOMICS: intrinsics
CHPL_GMP: gmp
CHPL_HWLOC: hwloc
CHPL_REGEXP: re2
CHPL_AUX_FILESYS: none
##########################

[Cleaning file /Users/buddha/github/buddha314/numsuch/test/NamedMatrix-test.chpl]
[Starting /Users/buddha/github/chapel/util/test/sub_clean NamedMatrix-test.chpl Tue Feb 06 14:03:49 PST 2018]
[Starting sub_clean - Tue Feb 06 14:03:49 PST 2018]
[pwd: /Users/buddha/github/buddha314/numsuch/test]
Cleaning test: NamedMatrix-test

[Working on file NamedMatrix-test.chpl]
[Starting /Users/buddha/github/chapel/util/test/sub_test Tue Feb 06 14:03:49 PST 2018]
[Starting subtest - Tue Feb 06 14:03:49 PST 2018]
[test: /Users/buddha/github/buddha314/numsuch/test/NamedMatrix-test.chpl]
[Executing compiler /Users/buddha/github/chapel/bin/darwin/chpl -o NamedMatrix-test --cc-warnings -M../src -M/Users/buddha/github/cdo/src -L/usr/local/Cellar/openblas/0.2.20/lib -lblas -I/usr/local/Cellar/openblas/0.2.20/include NamedMatrix-test.chpl < /dev/null]
[Elapsed compilation time for "/Users/buddha/github/buddha314/numsuch/test/NamedMatrix-test (compopts: 1)" - 14.918 seconds]
[Error cannot locate compiler output comparison file /Users/buddha/github/buddha314/numsuch/test/NamedMatrix-test.good]
[Compiler output was as follows:]
../src/MatrixOps.chpl:36: internal error: SCA0500 chpl version 1.17.0 pre-release (f732cb37da)
Note: This source location is a guess.

Internal errors indicate a bug in the Chapel compiler ("It's us, not you"),
and we're sorry for the hassle.  We would appreciate your reporting this bug --
please see https://chapel-lang.org/bugs.html for instructions.  In the meantime,
the filename + line number above may be useful in working around the issue.

[Elapsed time to compile and execute all versions of "/Users/buddha/github/buddha314/numsuch/test/NamedMatrix-test" - 14.919 seconds]
[Finished subtest "/Users/buddha/github/buddha314/numsuch/test/NamedMatrix-test" - 15.166 seconds]

[Removing /var/folders/g5/4zwyfk513hd94tjs1fm5mdq00000gn/T/chplTestTmpDir.f4HGaP directory]
[Done with tests - 180206.140404]
[Log file: /Users/buddha/github/chapel/test/Logs/buddha.darwin.log ]

[Test Summary - 180206.140404]
[Error cannot locate compiler output comparison file /Users/buddha/github/buddha314/numsuch/test/NamedMatrix-test.good]
[Summary: #Successes = 0 | #Failures = 1 | #Futures = 0 | #Warnings = 0 ]
[Summary: #Passing Suppressions = 0 | #Passing Futures = 0 ]
[END]

I'm attaching files for posterity. NO, I'm not. Github won't let me select .chpl files, so I'm pasting the relevant bits.

NamedMatrix.test.chpl

use NumSuch;

var nv: int = 8,
    D: domain(2) = {1..nv, 1..nv},
    SD: sparse subdomain(D),
    X: [SD] real;

var vn: [1..0] string;
vn.push_back("star lord");
vn.push_back("gamora");
vn.push_back("groot");
vn.push_back("drax");
vn.push_back("rocket");
vn.push_back("mantis");
vn.push_back("yondu");

SD += (1,2); X[1,2] = 1;
SD += (1,3); X[1,3] = 1;
SD += (1,4); X[1,4] = 1;
SD += (2,2); X[2,2] = 1;
SD += (2,4); X[2,4] = 1;
SD += (3,4); X[3,4] = 1;
SD += (4,5); X[4,5] = 1;
SD += (5,6); X[5,6] = 1;
SD += (6,7); X[6,7] = 1;
SD += (6,8); X[6,8] = 1;
SD += (7,8); X[7,8] = 1;

var nm = new NamedMatrix(X=X);
/* Should Error out, too few names */
try {
  nm.setRowNames(vn);
} catch {
  writeln("throwing to second!");
}

vn.push_back("nebula");
writeln(nm.setRowNames(vn));

MatrixOps.chpl (edited)

use Cdo,
    LinearAlgebra,
    LinearAlgebra.Sparse,
    Random;

class NamedMatrix {
  var D: domain(2),
      SD = CSRDomain(D),
      X: [SD] real,  // the actual data
      rowNames: [1..0] string;

   proc init(X) {
     this.D = {X.domain.dim(1), X.domain.dim(2)};
     super.init();
     this.loadX(X);
   }
}

proc NamedMatrix.loadX(X:[]) {
  for (i,j) in X.domain {
    this.SD += (i,j);
    this.X(i,j) = X(i,j);
  }
}

proc NamedMatrix.setRowNames(rn:[]): int throws {
  if rn.size != X.domain.dim(1).size then throw new Error();
  for i in 1..rn.size {
    this.rowNames.push_back(rn[i]);
  }
  return this.rowNames;
}

[buddha314]

Looks like the problem was that I was returning a string array when I said I was gonna return an int. That's fine, but the compiler error is weird or I read it wrong.

[lydia-duncan]

Looks like we're failing an assert in scalarReplace. Somehow a move or assign has been created over two things with different valTypes (when I would've expected that to be checked already). Maybe this is a promotion issue? I know very little about this pass, though, so that's just a guess

[buddha314]

I got it to compile using

proc NamedMatrix.setRowNames(rn:[]): string throws {
  if rn.size != X.domain.dim(1).size then throw new Error();
  for i in 1..rn.size {
    this.rowNames.push_back(rn[i]);
  }
  return this.rowNames;
}

Note the string in there, even though I am returning an array of strings. Is that supposed to be okay?

[lydia-duncan]

That doesn't seem like it should be okay to me, and does make me more inclined to blame promotion in some form.

[lydia-duncan]

Can you print out the return value? Does it give you a single string or the array?

[buddha314]

This is exactly the Peter Principle in action, I agree.

Return is now

throwing to second!
star lord gamora groot drax rocket mantis yondu nebula

[lydia-duncan]

Do you get the appropriate error message if you comment out the error handling?

[buddha314]

start_test gets mad if I do that

nm.setRowNames(vn);

> uncaught Error:
>   ../src/MatrixOps.chpl:42: thrown here
>   NamedMatrix-test.chpl:32: uncaught here

Is that what you wanted to see or something else?

[lydia-duncan]

I meant remove the throws declaration from the method and the throw new Error from within the body

[buddha314]

I believe you mean:

//proc NamedMatrix.setRowNames(rn:[]): string throws {
proc NamedMatrix.setRowNames(rn:[]): int {
  //if rn.size != X.domain.dim(1).size then throw new Error();
  for i in 1..rn.size {
    this.rowNames.push_back(rn[i]);
  }
  return this.rowNames;
}

which gives me the error

> ../src/MatrixOps.chpl:46: internal error: SCA0500 chpl version 1.17.0 pre-release (04472a0924)
> Note: This source location is a guess.
>
> Internal errors indicate a bug in the Chapel compiler ("It's us, not you"),
> and we're sorry for the hassle.  We would appreciate your reporting this bug --
> please see https://chapel-lang.org/bugs.html for instructions.  In the meantime,
> the filename + line number above may be useful in working around the issue.
>

This has the mismatch of return type.

[lydia-duncan]

Yes, that's what I meant. Looks like this is independent of error handling, so I'll update the issue description

[buddha314]

Go for it!

[buddha314]

I like "Odd behavior" better, sounds like you are introducing a novel.

[lydia-duncan]

Internal error is a bit more descriptive, though :)

[buddha314]

B O O O R R I N G

All, I am now officially BORED TO DEATH. YOU! YOU KILLED ME LYDIA!

[lydia-duncan]

Would you prefer "The curious case of the array returned from a function"?

[buddha314]

It IS a curious case, isn't it? How intriguing!

Do go on...