Look into implementing Recognisers.

[SirWumpus]

We come to consensus on a number of questions recently, and I just uploaded my reference implementation. I have checked, this implementation works in Post4 too.

@ruv is this the current recogniser proposal or are there more than one to consider?

[ruv]

Yes, this is the current proposal, along with Common terminology for recognizers. At present, the proposal has not yet been updated following recent discussions.

[SirWumpus]

We come to consensus on a number of questions recently, and I just uploaded my reference implementation. I have checked, this implementation works in Post4 too.

@ruv Did you do anything special to test this? I presume you overrode the internal REPL with the recogniser's.

[ruv]

Did you do anything special to test this?

There are tests for recognizers: name.test.fth, number.test.fth, word.test.fth.

They can be run with well-known ttester.fth.

I presume you overrode the internal REPL with the recogniser's.

In this implementation nothing is overridden (at the moment). It provides example.interpret ( i*x "ccc" -- j*x ) and example.evaluate ( i*x sd -- j*x ) in example.interpret.fth.

Usage example:

example.interpret 2 3 + .
s" 2 3 + . " example.evaluate

In this code, 2 3 + . is interpreted using the recognizer recognize-forth-default (that is assigned to the perceptor).

[SirWumpus]

SideScuttal:

They can be run with well-known ttester.fth.

I have my own version test/assert.p4 that supports the same words and adds a few more.