Skip to content

Latest commit

 

History

History
31 lines (25 loc) · 1.41 KB

DesignNotes-D.md

File metadata and controls

31 lines (25 loc) · 1.41 KB
Raw

Design Notes Part D -- Parsing Implementation

Late April 2024

OK, so Design Notes Part C sketched out a system for examining connector sets, and using LG style parsing to arrive at candidate linkage diagrams. This is followed by a step to select one linkage, and then a step to perform the actual hookups. The result is, ideally, a single network that, when executed, will run automatically.

The design has several unsolved issues.

  • It required the specification of MyFooAgent which contains the needed connector sets for that Agent. The current LookatLink requires this to be a C++ class. Clearly, this is impractical.

  • An automated way of making CrossSections is needed. Probably a CrossSection toolkit. It has to be Atomese, of course.

  • A HookupLink ...

  • Clear up the connector sex semantics. It's messy, right now. Clear up flow data types, they're wonky: either strings or ItemNodes or whatever. String processing is biting us in the butt.

  • The xterm demo creates hookups of streams running in inf loops, because OutputStrem::do_write_out() enters an inf loop when the source is a stream. The IRC demo does not run in an inf loop, because, after all the filtering, the code does not realize that there is a stream source at the far end, and as a result, each I/O has to be single-stepped. So which is it: automatic info loop, or single-stepping?