Nick Roberts wrote:

>> Hi,
>> the following patch removes 'type prefix' from output of
>> -var-evaluate-expression applied to objects of function type (not
>> pointers to functions, but functions).
>> 
>> It causes no regression in the testsuite for me.
> 
> But I guess a new test would help prevent a regression (in the code) in
> the future.

Yes, can you suggest which file should I add this new testcase too?


>> Changelog:
>> 
>> 2006-03-15 Vladimir Prus <ghost@cs.msu.su>
>>         
>>         * c-valprint.c
>>          (c_val_print): Don't print type prefix for functions.
>>          (c_value_print): Print type prefix for functions here.
> 
> 
>> Patch attached.
> 
> ...
>> +  if (TYPE_CODE (type) == TYPE_CODE_FUNC)
> 
> This condition must always be true here (case TYPE_CODE_FUNC:).

This chunk is in 'c_value_print', where there's no switch.      

> 
> 
> Where's the patch for c_value_print?

I guess I'd better send the patch with "-p". It's attached, hopefully it's
more clear.

And thanks for the hint about -p!

- Volodya