precedence of logical operators

[dram]

In DRM, & and | have same precedence. Open Dylan is confirm to DRM, but in Mindy & has higher precedence, as specified in compiler/info.c.

For example, Open Dylan and Mindy show different result for following code:

format-out("('a' | 'b') & 'c' => %s\n", ('a' | 'b') & 'c');
format-out("'a' | ('b' & 'c') => %s\n", 'a' | ('b' & 'c'));
format-out("'a' | 'b' & 'c' => %s\n", 'a' | 'b' & 'c');

Mindy:

('a' | 'b') & 'c' => c
'a' | ('b' & 'c') => a
'a' | 'b' & 'c' => a

Open Dylan:

('a' | 'b') & 'c' => c
'a' | ('b' & 'c') => a
'a' | 'b' & 'c' => c

Curious that why Dylan is designed to have same precedence for & and |, which is different to many other languages.

[waywardmonkeys]

I wonder if any Dylan code contained within Mindy relies upon this.

[wvdschel]

One way to find out?

On 29 October 2015 15:07:40 CET, Bruce Mitchener notifications@github.com wrote:

I wonder if any Dylan code contained within Mindy relies upon this.

---

Reply to this email directly or view it on GitHub: https://github.com/project-mindy/mindy/issues/39#issuecomment-152191641

Sent from my Android device with K-9 Mail. Please excuse my brevity.

[waywardmonkeys]

@dram Have you tried fixing this and seeing what happens?

[dram]

I'm a bit busy recently, do not have time to inspect whole Mindy codebase for usages of logical operators. If someone can help out, please go ahead.

[waywardmonkeys]

A quick grep finds:

libraries/regular-expressions/interface.dylan:      if (~substring-start | (count & (count <= string-number)))
libraries/string-extensions/character-type.dylan:  (c >= 'a' & c <= 'z')  |  (c >= 'A' & c <= 'Z');
libraries/string-extensions/character-type.dylan:  (c >= 'a' & c <= 'z')  |  (c >= 'A' & c <= 'Z')  |  (c >= '0' & c <= '9');
libraries/string-extensions/character-type.dylan:  (c >= '0' & c <= '9')  |  (c >= 'a' & c <= 'f')  |  (c >= 'A' & c <= 'F');
mindy/dbmc/header.dylan:                  | ('A' <= char & char <= 'Z')
mindy/dbmc/header.dylan:                  | ('0' <= char & char <= '9')
mindy/dbmc/header.dylan:              | ('A' <= char & char <= 'Z'))
mindy/dbmc/lexer.dylan:          elseif ((char == '\n' | char == '\r') & posn > lexer.line-start)
mindy/dbmc/lexer.dylan:          elseif ((char == '\n' | char == '\r') & posn > lexer.line-start)
mindy/dbmc/lexer.dylan:          elseif ((char == '\n' | char == '\r') & posn > lexer.line-start)
mindy/dbmc/lexer.dylan:          if ((char == '\n' | char == '\r') & posn > lexer.line-start)
tests/dylan-test.dylan:  (#t & #t)                        | signal("#t & #t is not true!\n");
tests/dylan-test.dylan:  (#f | #f)                        & signal("#f | #f is not false!\n");
to-be-migrated/common-dylan/format.dylan:            | (str[start] = '0' & str[start + 1] = 'x')))
to-be-migrated/file-system/helpers.dylan:  size(ans) > 0 & (ans[0] = 'Y' | ans[0] = 'y')
to-be-migrated/locators/locators.dylan:            | (base & base ~== $unsupplied)
to-be-migrated/locators/locators.dylan:            | (extension & extension ~== $unsupplied))
to-be-migrated/tests/streams.dylan:                  & (~test-function-direction | test-function-direction == #"input");
to-be-migrated/tests/streams.dylan:                  & (~test-function-direction | test-function-direction == #"output");
tools/melange/c-decl.dylan:  decl.map-type | (~explicit-only? & decl.type-name);
tools/melange/c-decl.dylan:      | (~explicit-only? & decl.type-name);
tools/melange/c-decl.dylan:    decl.map-type | (~explicit-only? & decl.type-name);
tools/melange/c-decl.dylan:    | (~explicit-only? & decl.type-name);
tools/melange/c-decl.dylan:  decl.map-type | decl.type.map-type | (~explicit-only? & decl.type-name);
tools/melange/c-decl.dylan:    | (~explicit-only? & decl.type-name);
tools/melange/c-lexer-cpp.dylan:          & (empty?(rest) | head(rest) == #"accept"))
tools/melange/c-lexer-cpp.dylan:          & (empty?(rest) | head(rest) == #"accept")
tools/melange/c-lexer.dylan:             | ~(whitespace?(contents[i]) & contents[i] ~== '\n')))
tools/melange/c-lexer.dylan:    if (pos = contents.size | (cpp-line & contents[pos] == '\n'))

I think that it is safe to fix this bug.

[dram]

Just find that actually both precedence and association are incorrect. This bug is quite anonying, hope that no code rely on this.