What about merge function gen_trace_for_expr and function gen_eval_for_expr together?

What about merge function gen_trace_for_expr and function gen_eval_for_expr together?

[Hui Zhu]

Hi guys,

When I read the code, I found that most part of these functions are same:
2c2
< gen_trace_for_expr (CORE_ADDR scope, struct expression *expr)
---
> gen_eval_for_expr (CORE_ADDR scope, struct expression *expr)
12c12
<   trace_kludge = 1;
---
>   trace_kludge = 0;
16,17c16
<   /* Make sure we record the final object, and get rid of it.  */
<   gen_traced_pop (expr->gdbarch, ax, &value);
---
>   require_rvalue (ax, &value);


What about add a new argument for example trace or eval and merge them together?

Thanks,
Hui

Re: What about merge function gen_trace_for_expr and function gen_eval_for_expr together?

[Stan Shebs]

On 6/11/12 5:15 PM, Hui Zhu wrote:
> Hi guys,
>
> When I read the code, I found that most part of these functions are same:
> 2c2
> < gen_trace_for_expr (CORE_ADDR scope, struct expression *expr)
> ---
>> gen_eval_for_expr (CORE_ADDR scope, struct expression *expr)
> 12c12
> <   trace_kludge = 1;
> ---
>>   trace_kludge = 0;
> 16,17c16
> <   /* Make sure we record the final object, and get rid of it.  */
> <   gen_traced_pop (expr->gdbarch, ax, &value);
> ---
>>   require_rvalue (ax, &value);
>
>
> What about add a new argument for example trace or eval and merge them 
> together?

I certainly considered that when introducing gen_eval_for_expr, but it 
seemed like repeatedly testing a flag for the parts that are different 
was going to make it more complicated than having the two separate 
functions.  I also expected that they might need to diverge a bit more 
over time, for instance if trace_kludge were to be gotten rid of (as it 
should be).

Stan
stan@codesourcery.com