Re: MPS image cache

[Gerd Möllmann]

Eli Zaretskii <eliz@gnu.org> writes:

>> Apparently it only works as is on 64-bits and so on.
>
> How can you be sure the compiler packs the struct fields, even on a
> 64-bit platform?  If the fields are not packed, the struct could take
> more than 1 word on a 64-bit platform as well.  Compilers are known to
> align struct members for faster access.

It is on my platform.

>> What do you expect me to do?
>
> Can you explain what the code is supposed to do, and why you need that
> assertion?  

I made the compiler my check my assumption at compile time. I find that
pretty cool of me.

> "we need to know how large the vector is to scan it" didn't explain it
> enough for me to understand.

We allocate a vector of N pointers. The N is found in the header when we
scan it. It's the size of the MPS object we scan minus the header
divided by how large a pointer is.
>
>> A thing I promise not to do is to develop and check on any platform
>> except mine.
>
> I thought I was helping to test on a platform where you don't, but I
> need help in understanding the code, so I could think about possible
> ways of fixing it.  We are not abandoning 32-bit platforms any time
> soon, so a solution for this will be needed.

Yeah, I know, sorry. My sciatica is killing me, and I'm probably a bit
hard to handle these days :-(