Re: posix_fadvise v22

I'll send another path with at least 1 and 3 fixed and hunt around
again for a header file to put this guc into.

On 10 Dec 2008, at 04:22, ITAGAKI Takahiro <itagaki(dot)takahiro(at)oss(dot)ntt(dot)co(dot)jp
> wrote:

> Hello,
>
> Gregory Stark <stark(at)enterprisedb(dot)com> wrote:
>> Here's an update to eliminate two small bitrot conflicts.
>
> I read your patch with interest, but found some trivial bad manners.
>
> * LET_OS_MANAGE_FILESIZE is already obsoleted.
> You don't have to cope with the option.

Huh I didn't realize that. I guess the idea is that users just
configure a very large segment size to get the old behaviour.

>
> * Type mismatch in prefetch_pages
> A variable prefetch_pages is defined as "unsigned" or "int"
> in some places. Why don't you define it only once in a header
> and include the header in source files?

Just... Which header?

> * Assignment to prefetch_pages
> What do "+0.99" means here?
> [assign_io_concurrency()]
> + prefetch_pages = new_prefetch_pages+0.99;
> You want to do as follows, right?
> + prefetch_pages = (int) ceil(new_prefetch_pages);

Sure