Never underestimate the usefulness of human-readable data. (was Re: [uf-discuss] Non-HTTP/HTML microformats)

[Tantek Ç elik]

On 11/30/05 4:55 PM, "Ryan King" <ryan at technorati.com> wrote:

> On Nov 30, 2005, at 3:22 PM, Scott Reynen wrote:
> 
>> Ryan King wrote:
>> 
>>> You see, one of the beauties of The Unix Way, is that, just like
>>> microformats, you can use the same output for both humans and
>>> machines. For example, `ls .` can be read by a human or piped into
>>> something else with `ls . | foo`. No need for two separate
>>> formats. (note to self: need to write blog post comparing µF's to
>>> The Unix Way.)
>> 
>> In addition to removing repetition, using a more human-readable
>> format for machine-only communication greatly simplifies
>> debugging.  That's why JavaScript is far more popular than assembly
>> language.
> 
> And developing in general. For example, I write lots of shell scripts
> that end up looking something like:
> 
> cat *.log | cut -f2 | grep foo | perl -e "while(<>){print doSomething
> ($_);}" | sort | uniq -c
> 
> Of course, to create that, I do it one step at a time and inspect the
> output (often using head or tail instead of cat).
> 
> Never underestimate the usefulness of human-readable data.

Ryan, this is an excellent point, and I feel like something that needs to be
added to that other blog-post-in-progress based on the last discussion of
"The tools will save us" (NOT!) that we had on the list.

There is SO MUCH evidence (like the examples you just gave) for the
usefulness of making data formats be human-readable that it really makes you
wonder why so many otherwise intelligent people keep wanting to do
otherwise.

Perhaps it is time we started separate pages for each of the microformats
principles on the wiki, and provided deeper explanations and examples for
each.

Thanks,

Tantek