Mail Archives: geda-user/2015/02/08/12:27:34
On Feb 8, 2015, at 9:32 AM, Chris Smith <space DOT dandy AT icloud DOT com> wrote:
>>
>> The file format should be language-neutral. We have that now.
>
> No we don't. We have a format that's clearly biased towards your toolset.
It’s biased toward parsing with sscanf() and similar functions. This is a common, familiar approach implemented by many toolsets. That is not, however, what I usually use. For my purposes I would prefer a format that prefixed *every* line with a tag and used tab as a delimiter. But what we have is good enough.
Canonicalized free form is bad: you wind up with writers whose authors don’t understand all of the rules (or think they’re stupid), and then not all readers can cope. An example is FITS: headers look like they allow fields to start in arbitrary columns but they are actually don't. I once wrote a FITS library that was extremely liberal about what it read and extremely rigorous about what it wrote. It major use wound up being a single program, fitscanon, that would read and rewrite a nonstandard file so that other parsers could read them.
John Doty Noqsi Aerospace, Ltd.
http://www.noqsi.com/
jpd AT noqsi DOT com
- Raw text -