Re: Command-line program to convert 'human' sizes?

[Pádraig Brady]

On 12/05/2012 12:56 AM, Assaf Gordon wrote:
> Pádraig Brady wrote, On 12/04/2012 07:31 PM:
> > On 12/05/2012 12:19 AM, Jim Meyering wrote:
> > > Pádraig Brady wrote:
> > > > On 12/04/2012 11:35 PM, Assaf Gordon wrote:
> > > > > Pádraig Brady wrote, On 12/04/2012 06:11 PM:
> > > > > > On 12/04/2012 10:55 PM, Assaf Gordon wrote:
> > > > > > > > Pádraig Brady wrote, On 12/04/2012 11:30 AM:
>
> < snip long discussion >
>
> Would the following be acceptable:
> 1. remove "--to=<NUMBER>" option

I wish I could find my notes on this.
Yes ignore for now. I might remember what
I was thinking here later.

> 2. surplus characters following immediately after converted number trigger a 
> warning (error?),
>    except if the following characters match exactly the "suffix" parameter.

Right. Might as well error as give a warning?

> Regarding "--format":
> The implementation doesn't really use "printf", so "%d" isn't directly usable.

So we lose zero padding, base conversion, etc. but...

> One option is to tell the user to use "%s" (instead of "%d"), and we'll simply put the 
> result of "human_readable()" as the string parameter in vasnprintf - this will be flexible in terms 
> of alignment.
> Another option is the remove "--format" option, and replace it with "--padding" 
> or similar.


> Regarding grouping (thousands separator):
> This only has an effect when no using "--to=SI" or "--to=IEC", right?
> Perhaps we can add a separate option "--grouping", and simply turn on the 
> "human_grouping" flag? (easy to implement).

... such features may be better with explicit options
for --padding, --grouping, --base.
Yes grouping wouldn't make much sense with --to=SI

cheers,
Pádraig.