Quoth Sebastian Spaeth on Aug 12 at 10:07 am:
> On Mon, 11 Jul 2011 17:05:32 -0400, Austin Clements <amdragon@MIT.EDU> wrote:
> > So what would be a good format?  One possibility would be to
> > NULL-delimit the query part; as distasteful as I find that, this part
> > of the search output isn't meant for user consumption.  Though I fear
> > this is endemic to the dual role the search output currently plays as
> > both user and computer readable.
> 
> Perhaps we take a queue from xargs and have a command line switch for \n
> vs \0 output?
> 
> --null
> -- 0
>   Input items are terminated by a null character instead of by
>   whitespace, and the quotes and backslash are not special (every
>   character is taken lit- erally).

This was one of the approaches I considered, but given that JSON
parsing (with my optimized json.el) is nearly as fast as regexp-based
parsing (which would also be the fastest way to parse \0-separated
output) and is flexible, robust, and structured (unlike any simple
delimited text format), I concluded we should just go with JSON.  If
we care about speed enough to introduce another format, I'd propose
S-expressions over a new text format, since they double parsing
performance compared to text, have the same structural benefits as
JSON, and JSON and S-expressions could even share the same formatting
code (so there's no chance of representation divergence).

BTW, reviving the JSON search parser is on my shortlist.  I should
also bundle up my optimized json.el, since it should seriously boost
the performance of notmuch-show, too.