Oh, neat! Can you share the code? Hmm. I hadn't thought about how to handle the interpretation of nouns, or the extension of that syntax. For non-nouns, I was going to let users override the definitions of J primitives using my cover-function-substitution utilities.
-Dan -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Don Guinn Sent: Saturday, February 12, 2011 7:53 PM To: Programming forum Subject: Re: [Jprogramming] JUICE Years ago I got interested in quaternions and wrote a tool to calculate quaternions in J. The tools on the web were just too hard to use. It wasn't just a quaternion calculator, it would read a J sentence and using the J parser allow me to extend J numbers into quaternions and replace J primitives with names which would handle quaternions. For example in the ijx window. Normal J sentences except the numbers are quaternions: ]q=.2i3j4 1+/2k1 1i2j3k4 5 4i3j4k1 3i5j7k4 7i3j4 3k1 2i2j3k4 6 +:q 8i6j8k2 6i10j14k8 14i6j8 6k2 4i4j6k8 12 All I really cared about was handling primitive verbs and nouns. Later to really learn J better I wrote a tool called MN for mnemonics (improperly named) to translate primitives into the appropriate name for the primitive, allowing for parity and position in a sentence. In this case I was interested in handling adverbs and conjunctions and other things in a J sentence properly. A much more involved problem. The parsing started getting out of hand. Still haven't finished it satisfactorily. So I have been thinking of trying to use trace so I could get inside the parsing better. On Sat, Feb 12, 2011 at 5:30 PM, Dan Bron <[email protected]> wrote: > Don wrote: > > Look at the verb "trace" defined in > > J7 '~addons/general/misc/trace.ijs'. > > Thanks. I thought about this too, but figured if I could just write an > appropriate x&;: and uni2Puny (which I'd have to do anyway), and applied ". > , that would be enough. That is, I wouldn't have to write my own parser, or > understand or adapt the one in trace. > > > I should have gone to the trouble to understand > > trace as it would have done a much better job. > > But I'm interested in why you concluded this. Can you give me a little > background on the project and why you think trace would have done a better > job? > > > handling copulas > > I can see how writing my own parser would allow me to support sentences > like: > > ? =: =: > > ... but I'm not sure that's worth it to me. I don't prefer it, but I think > I'm ok with forcing my users to write copulae the standard way. > > -Dan > > > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
