maria-developers team mailing list archive

[Sergei Golubchik <serg@xxxxxxxxxxx>]

Hi, Aleksey!

On Jun 03, Aleksey Midenkov wrote:
> Hi Sergei!
> 
> Do you plan to fix everything you just said and push yourself or let
> me push it as is?

Neither. As a reviewer I plan to reply "ok to push" when the code is
changed according to the review.

> Btw, there were many allocations over one loop: that's why I used
> mem_root. multi_malloc() won't do the job there. That could be done
> either with explicit moving pointer or some utility class holding such
> pointer. I decided to upgrade mem_root for such pattern so it could do
> the double job and we had a cleaner code.

Yes, I mean

  my_multi_malloc(MYF(MY_WME),
    &m_ordered_rec_buffer, alloc_len,
    &blob_storage, table->s->blob_fields * sizeof(Ordered_blob_storage *),
    &ptr, table->s->blob_fields * sizeof(Ordered_blob_storage), NULL);

  for (uint j= 0; j < table->s->blob_fields; ++j, ++ptr)
    blob_storage[j]= new (ptr) Ordered_blob_storage;

Regards,
Sergei
VP of MariaDB Server Engineering
and security@xxxxxxxxxxx