Moses work 2022-10-04 to 2022-10-07
to : Martin Holmes
Minutes: 630
Finally getting a chance to look at the feature requests and bug reports from EH’s work earlier in the summer, I debugged and fixed several issues in the PDF, and had some useful discussions with SK about other features. There is a need to provide initial-letter links in the online dictionary for the list of entries, but it’s actually not going to be trivial because an initial letter is not necessarily a single letter but may be a sequence of two or three glyphs. At first thought, I was wondering if I should use XSL grouping, but that won’t be easy; instead, it might be simpler to detect any change in the initial three letters and test whether it matches a significant transition, but even that is hard. I think the way to do it might be reduce each pair of words using a greedy regex to only the initial component that matches one of our significant initial sequences, and then those two initial sequences are not identical we have a transition. Anyway, it’ll be an interesting challenge.
Meanwhile, following a long and useful meeting on Thursday, I implemented auto-orthing in the online dictionary in preparation for adding in the letter transition anchors/links, and I’ve sketched out the algorithm for the above function. We’ve decided on a number of small modifications to the print dictionary, which I’ll implement after that.
On Friday, I wrote the algorithm for determining whether a transition has happened between one initial letter and the next, and wrote a stack of XSpec tests to debug and finalize it. Then I implemented the initial letter headings and links at the top of the page. I migrated the same feature to the English wordlist, and in the process fixed a couple of other bugs and issues. The online dictionary is looking a bit better as a result of this.