SASADA Koichi <ko1 atdot.net> writes:

>> Another bug just found:
>> 
>> #583<3|>lilith:~/src/yarv$ ./ruby -e 'a = ->(foo){ p foo }'
>> BUG: unknown node (default): NODE_LAMBDA
>> :0:in `initialize': compile error (SyntaxError)
>
> Matz decided that "->" is obsolete.

...and there was much rejoicing. ;-)

>> (gdb) r foo
>> Starting program: /Users/chris/src/yarv/ruby foo
>> Reading symbols for shared libraries ... done
>> 
>> Program received signal EXC_BAD_ACCESS, Could not access memory.
>> frame_func_id (cfp=0x485000) at eval.c:1869
>> 1869        yarv_iseq_t *iseq = cfp->iseq;        
>> (gdb) bt
>> #0  frame_func_id (cfp=0x485000) at eval.c:1869
>> #1  0x00002e18 in error_pos () at eval_error.h:42
>> #2  0x0000309c in error_print () at eval_error.h:108
>> #3  0x00003624 in error_handle (ex=823640) at eval_error.h:250
>> #4  0x00005a54 in ruby_options (argc=2, argv=0xbffffc38) at eval.c:175
>> #5  0x00002c58 in main (argc=2, argv=0xbffffc38, envp=0x20000000) at main.c:45
>
> I can't recreate your bug.
>
>> # ./ruby foo.rb
>> ./ruby: No such file or directory -- foo.rb (LoadError)
> at my environment (on x86 linux)

Here is an easy fix, I think it should not affect anything (can
nd_line ever return 0 intentionally?)


Index: eval_error.h
===================================================================
--- eval_error.h        (revision 429)
+++ eval_error.h        (working copy)
@@ -39,13 +39,13 @@
 {
     ruby_set_current_source();
     if (ruby_sourcefile) {
-       if (rb_frame_callee()) {
+       if (ruby_sourceline == 0) {
+           warn_printf("%s", ruby_sourcefile);
+       }
+       else if (rb_frame_callee()) {
            warn_printf("%s:%d:in `%s'", ruby_sourcefile, ruby_sourceline,
                        rb_id2name(rb_frame_callee()));
        }
-       else if (ruby_sourceline == 0) {
-           warn_printf("%s", ruby_sourcefile);
-       }
        else {
            warn_printf("%s:%d", ruby_sourcefile, ruby_sourceline);
        }


>> Another thing, but this probably is intented for 1.9, is that [*foo]
>> doesn't work for arrayification anymore.  What's the best way to do it
>> in 1.9?
>
> a = [1, 2]
> p [*a]     #=> [1, 2]
> a = 1
> p [*a]     #=> can't convert Fixnum into Array (TypeError)
>
> You means second error?  It's new specification on Ruby 1.9.  ... I have
> no idea about arrayification (it's first time I see this word) with 1.9.

Well, since 1.9 also drops default Object#to_a, I found no other easy way
to make an object an array unless it already is one.

> // SASADA Koichi at atdot dot net
-- 
Christian Neukirchen  <chneukirchen gmail.com>  http://chneukirchen.org

-- 
ML: yarv-dev-en quickml.atdot.net
Info: http://www.atdot.net/~ko1/quickml