[PATCH] Dynamic growable arrays for internal use

[Florian Weimer]

On 06/02/2017 09:37 AM, Carlos O'Donell wrote:
> In summary the remaining points look like this:
> 
> * Add rationale to malloc/dynarray.h as a comment explaining when and
>   why you would use it instead of a scratch buffer. What you write below
>   with some adjustments is good.

Okay, I added comments to dynarray.h.

>> diff --git a/support/capture_subprocess.h b/support/capture_subprocess.h
>> new file mode 100644
>> index 0000000..3265d63
>> --- /dev/null
>> +++ b/support/capture_subprocess.h

>> +/* Invoke CALLBACK (CLOSURE) in a subprocess and capture standard
>> +   output, standard error, and the exit status.  The out.buffer and
>> +   err.buffer members in the result are null-terminated strings which
>> +   can be examined by te caller.  (out.out and err.out are NULL.)  */
> 
> s/te/the/g.
> 
> Suggest:
> can be examined by the caller (out.out and err.out are NULL).

Thanks, fixed.

>> +
>> +#include <support/test-driver.c>
>> diff --git a/support/xdup2.c b/support/xdup2.c
>> new file mode 100644
>> index 0000000..f4a3759
>> --- /dev/null
>> +++ b/support/xdup2.c
>> @@ -0,0 +1,28 @@
>> +/* pipe with error checking.
> 
> s/pipe/dup2/g

Fixed.

I adjusted the patch to the presence of check_mul_overflow_size_t in
<malloc-internal.h> and committed it.

Florian