Re: [PATCH 4/4] gdb: fix value_subscript when array upper bound is not known

[Simon Marchi via Gdb-patches <gdb-patches@sourceware.org>]

On 2020-12-07 10:14 p.m., Joel Brobecker wrote:
> [just for the sake of discussing generalities, rather than suggesting
> modifications to your patch]
> 
>>> Now, the bigger question is: Why are we even getting to this point?
>>> Shouldn't the bounds simply be resolved before we do the subscripting?
>>
>> If the array is c-style, there's no concept of high bound, so I don't think
>> there's a point resolving the high bound then.
> 
> I believe you sometimes do, for intance in...
> 
>     char something[5];

Right, but the language doesn't do bound checks, so I don't think the debugger
should either.  What I meant is that there's no need to resolve the upper bound
for c-style arrays, because we don't check the upper bound.

>> Again, I've never really been a user of such a language.  Intuitively, I'd
>> like the debugger to work as close to the source language as possible, so
>> I'd like it to validate the bounds in my expression.  But it might be that
>> people who use these languages sometimes need to side-step the language's
>> protections.
> 
> Speaking for myself, sometimes I indeed want to go beyond.
> 
> Even in C, when we have structures looking like this (don't know if we
> have any of those in our code anymore):
> 
>     struct something
>     {
>       size_t n_elems;
>       int buf[0];
>     };
> 
> You'd know that you need to go beyond the first element of the array
> to get to the rest of the data...

I would imagine that with a language that is more strict and does bound-checks
when accessing arrays, this kind of trick wouldn't exist.  So there probably
wouldn't be legitimate use cases for doing out-of-bounds accesses.

Simon