1 Feb 2006 17:11
Human-readable data dumps, was Re: optional subexp..
Bob the Hamster <Bob <at> HamsterRepublic.com>
2006-02-01 16:11:44 GMT
2006-02-01 16:11:44 GMT
On Tue, Jan 31, 2006 at 04:50:12PM -0500, Mike Caron wrote: > On 1/31/06, David Gowers <00ai99 at gmail.com> wrote: > > I implemented that regexp and finished the generic importing framework, **clip** > > Hmm, I almost like what I see. Almost. > > That looks like an array format (probably Python), except for the > bitsets, which I'd bet aren't. That would make parsing it difficult. > Can that be imported in some way? > > In any case, why not use XML*? It's perfect for this type of thing: > > * It's a standard format, supported by the W3C (albeit, intended for > online applications) > * It's human readable > * It's flexible > * It's easily parsable (especially since there are a number of > libraries to do it for you. I suspect python has an XML module, even) > * It can be formatted in any way you want, esp. via XSLT (if you don't > know what I mean, take a look at the plotdict.xml source, versus what > it looks like in the browser) > > Either way, a non-binary format of any kind would be useful, > especially for external tools. Imagine a world where a tool didn't > need to worry about internal formats, or version changes, or which > order the bytes go in, or whatever. They can parse the XML... er, I > mean, whatever format is adopted, and modify that. > > In summary, XML is good. Er... I mean, a text format of some kind would be good.(Continue reading)
RSS Feed