Moses 2023-05-01 to 2023-05-05
to : Martin Holmes
Minutes: 350
Getting back to work on the project, I picked up the two tasks related to number stems assigned by SK; the first was a bug in the root-based-index generation code, which I found and fixed; the second is a request to tweak a diagnostic which (I think) is already too complicated, so I’ve suggested that we split it out into two or three to keep it faster and simpler.
On Wednesday, I replaced the original diagnostic with three simpler ones, and these are mostly working except for one bug which I have to fix in the discovery of missing root symbols; this looks like it might be caused by a bug in Saxon’s optimizer, but I’m not sure of that yet.
On Thursday, I split out that complicated XPath into a choose element, and as expected the problem disappeared and the remaining cases were found.
On Friday, worked on the one remaining orthing rule, which turned out to be a bit complicated
because it applied to roots and stems, which don’t have <hyph>s. Solved
the problem by generating a <hyph> from the p-pron and feeding that
into the system, which seems to have worked.