On Mon, Aug 26, 2013 at 07:29:24AM -0600, Theo de Raadt wrote:
> > On Mon, Aug 26, 2013 at 11:27:36AM +0000, Stuart Henderson wrote:
> > > you will then end up with some of them switching to dreadful MiB etc. ;)
> > 
> > Are there strong opinions against following standards and start
> > converting to the proper terms for gigabytes (decimal, base 10, 1GB =
> > 1000^3 bytes) and gibibytes (binary, base 2, 1 GiB = 1024^3 bytes)?
> 
> I have a strong opinion against it.  The benefits are overstated.

Dammit, I'll create a fork called BiKiNiBSD! ;)
 
> I believe "When in Unix, ..."
> 
> > After all it's been a while since it was logical (!) to infer that since
> > 1024^1 (kibi) is *almost* 1000^1 (kilo), then 1024^3 (gibi) must
> > *almost* be 1000^3 (giga).. ;)
> 
> "Almost" has never been the problem in real life, or to bring it
> closer to home -- in real system administration.  The question we face
> is "enough".  As the capacity of equipment grows faster than the needs
> for data storage, "enough" is easy to satisfy.

Agreed. Of course. My concern is not about tweaking space.

> These new units create as much harm by requiring more comprehension
> space -- ie. concern is not just about having the space, but you need
> to remember which of the two units you are in, so that you don't
> accidentally convert backwards once in a while.  It becomes critical
> to show the actual unit visibly, in every step, like in a column or on
> a label.
> 
> That is a step backwards.
> 
> Disk: wd0       geometry: 155061/16/63 [156301488 Sectors]
> Offset: 0       Signature: 0xAA55
>             Starting         Ending         LBA Info:
>  #: id      C   H   S -      C   H   S [       start:        size ]
> -------------------------------------------------------------------------------
>  0: 00      0   0   0 -      0   0   0 [           0:           0 ] unused    
>   
>  1: 00      0   0   0 -      0   0   0 [           0:           0 ] unused    
>   
>  2: 00      0   0   0 -      0   0   0 [           0:           0 ] unused    
>   
> *3: A6      0   1   1 -  16382  15  63 [          63:    16514001 ] OpenBSD   
>   

(So it is true about the backdoors in OpenBSD? How else did you get
access to my disk!?!)

Okay, as for any need to understand how (why) disk space should (must)
be understood in terms of base-2 sizes, I'm really way out of my league.

Lets say I'm happening to have lots of smaller disks that I'd like to
create partitions for on larger disks. Reading on the label on one such
small disk that it has a capacity of 160GB, and knowing that this means
160 * 1000^3 bytes, makes it easy to create a partition that big on a
larger disk without having to remember the 9 or 10 digit sector size or
to look up the size in GB (eg. GiB) on the MBR or disklabel for the
smaller disk. Maybe a stupid example, perhaps, but still. It's just
about the potential mess with confusing unit values.

I guess all it boils down to is the question why OpenBSD shouldn't use
standard unit names, that is GiB for gigabytes and GB for gibibytes?

http://en.wikipedia.org/wiki/Gigabyte
http://en.wikipedia.org/wiki/Orders_of_magnitude_(data

Wouldn't it be according to OpenBSD's goal to follow standards, avoiding
ambiguities? I don't think users will have problems remembering what
units they are in, at least not if the units are correct.

> If the argument was about "almost enough space", now it is about "not
> enough" space to put the unit letters.
> 
> > Personally I would love to see disklabel(8) default to display sizes in
> > base-10 and with something like an optional -i or -2 switch to display
> > information in the old (current) base 2 definition. At least it would be
> > nice if it were using proper units - like GiB instead of GB.
> 
> And then we can modify the installer to ask which unit people prefer?
> 
> Not a step forward.

Forget that I asked about having optional switches. However, I do feel
that my above question about using proper (!) unit names is relevant.
It's not a big issue and I may not be "binary" enough to understand the
importance of these matters.

Cheers,

Erling

Reply via email to