[ruby-core:123258] [Ruby Feature#21573] Simpler syntax errors

Issue #21573 has been updated by mame (Yusuke Endoh).


Before hastily adding modes, we should clarify the problem.

In the proposed patch, the error message will be as follows:

```
$ echo -e "1 +x 2 do\ndo" | SIMPLE_ERRORS=1 ./miniruby -wc
-:1: warning: '+' after local variable or literal is interpreted as binary operator even though it seems like unary operator
-:1: warning: possibly useless use of + in void context
-:1: warning: possibly useless use of a literal in void context
./miniruby: -:1: syntax errors found (SyntaxError)
-:1: unexpected integer, expecting end-of-input
-:1: unexpected 'do', expecting end-of-input
-:1: unexpected 'do', ignoring it
-:2: unexpected 'do', ignoring it
```

Will this actually solve the problem?

* [The author of the issue states "and counting spaces to ascertain the column)"](https://github.com/ruby/prism/issues/3455#issue-2825415997). This suggests a desire to obtain the error column information. The proposed patch does not achieve this goal.
* [The author of the issue also states "Ruby can print one syntax error per line"](https://github.com/flycheck/flycheck/pull/2111#issuecomment-2628824273). But the proposed patch appears to display `-:1: unexpected 'do', expecting end-of-input` and `-:1: unexpected 'do', ignoring it`. Does this achieve the goal?
* [The triggering issue](https://github.com/flycheck/flycheck/pull/2111) was that the interpreter path is displayed since Ruby 3.4. This is simply a Ruby 3.3 incompatibility and should be fixe if possible?

As discussed in the ticket https://github.com/ruby/prism/issues/3455, it seems reasonable for tools like flycheck to switch to `Prism.parse_stream(STDIN).errors/warnings`.

----------------------------------------
Feature #21573: Simpler syntax errors
https://bugs.ruby-lang.org/issues/21573#change-114624

* Author: kddnewton (Kevin Newton)
* Status: Open
----------------------------------------
Right now both the prism and parse.y parsers add some context to syntax errors to make it easier to understand where they are coming from. This works well for humans, but is tough for tools (see https://github.com/ruby/prism/issues/3455 for some additional context). I would like to propose a feature that can be enabled via environment variable (or some other mechanism like a configure switch) that makes syntax errors output minimal information (i.e., `file:line:col: message`).

The diff would be fairly small for prism, below is the entirety of the change:

```diff
diff --git a/prism_compile.c b/prism_compile.c
index 578e6f240f..d14c23eb94 100644
--- a/prism_compile.c
+++ b/prism_compile.c
@@ -11018,13 +11018,15 @@ pm_parse_process_error(const pm_parse_result_t *result)
         (parser->error_list.size > 1) ? "s" : ""
     );
 
-    if (valid_utf8) {
+    if (!getenv("SIMPLE_ERRORS") && valid_utf8) {
         pm_parse_errors_format(parser, &parser->error_list, &buffer, highlight, true);
     }
     else {
         for (const pm_diagnostic_t *error = head; error != NULL; error = (const pm_diagnostic_t *) error->node.next) {
             if (error != head) pm_buffer_append_byte(&buffer, '\n');
-            pm_buffer_append_format(&buffer, "%.*s:%" PRIi32 ": %s", (int) pm_string_length(filepath), pm_string_source(filepath), (int32_t) pm_location_line_number(parser, &error->location), error->message);
+
+            pm_line_column_t location = pm_newline_list_line_column(&parser->newline_list, error->location.start, parser->start_line);
+            pm_buffer_append_format(&buffer, "%.*s:%" PRIi32 ":%" PRIu32 ": %s", (int) pm_string_length(filepath), pm_string_source(filepath), location.line, location.column, error->message);
         }
     }
```



-- 
https://bugs.ruby-lang.org/
______________________________________________
 ruby-core mailing list -- [e-mail]
 To unsubscribe send an email to [e-mail]
 ruby-core info -- https://ml.ruby-lang.org/mailman3/lists/ruby-core.ml.ruby-lang.org/