Re: 2nd attempt at Re: The right way(tm) of writing toString() (Was: Re: [PATCH] Field position attribute handling)

[Etienne Gagnon]

Hi Dalibor,

I would drop h) altogether.

Rationale: while a "naive" Java compiler would produce a series of iload
instructions, for reusing a local variable, an optimizing compiler or a jit
would completely remove this.  So, yes, the code that reuse a local could be
a little slower on a interpreter (I am sure that on sablevm this would be
negligible as iload is a very cheap operation), but good software engineering
is the goal here.

Not only do I personally find the following more readable, but it is also
easier to debug, as a debugger would trace through the instructions one by one:

public String toString()
{
  StringBuffer sb = StringBuffer();

  sb.append("Test[field_1=");
  sb.append(field_1);
  sb.append(",field_2=");
  sb.append(field_2);
  sb.append("]");

  return sb.toString();
}

The performance difference (except on "naive" interpreters) is negligible.

Note about performance:  The reason iload is expensive on naive interpreters,
is the cost of dispatching an additional instruction in the interpreter.
SableVM's inline-threaded engine does eliminate dispatch between successive
instructions of a basic block, so the cost of an additional iload is simply
that of the "iload" functionality, not that of (iload + dispatch) as found
in normal switch-based interpreters and direct-threaded interpreters.

So, in summary: it is best to leave such low-level optimizations in the hand
of the compiler, not of the developer.  The developer should instead care about
code readability and debuggability (i.e. software engineering issues).

Etienne

Dalibor Topic wrote:
> h) re-use the StringBuffer on the stack
>
>   Rationale: In order to write as efficient code using StringBuffer as 
> the good Java compilers, we need to reuse the StringBuffer instance on 
> the stack. So instead of writing multiple append statements, try to use 
> a single statement to perform all the work on the StringBuffer instance, 
> including creation, calling append, and calling toString() on it.
>
> Example code:
>
> public class Test{
>   private String field_1;
>   private int field_2;
>
>   public String toString() {
>     return
>     (new StringBuffer("Test[field_1=")).append(field_1)
>     .append(",field_2=").append(field_2)
>     .append(']')
>     .toString();
>   }
> }
>
> comments are welcome, as usual.
>
> cheers,
> dalibor topic


--
Etienne M. Gagnon, Ph.D.             http://www.info.uqam.ca/~egagnon/
SableVM:                                       http://www.sablevm.org/
SableCC:                                       http://www.sablecc.org/