Moses 2025-06-30 to 2025-07-04
to : Martin Holmes
Minutes: 565
On Monday, started on the latest TODOs for the print dictionary, and solved some,
while pushing the dialog forward on others. The last remaining thorny issue is
related to significant versus non-significant whitespace in <hyph>s,
and I’m still struggling with that. The process of creating the translated hyph is now
so arcane and complicated that I can’t see where the extra spaces are being introduced.
It’s also extemely difficult to set up tests without creating an entire internally-coherent
miniature data-set, and even if I did that, the chances that the obscure occasionally-appearing
whitespace would show up in that dataset are small.
On Wednesday, got back to the unwanted space issue in translated hyphs and added yet more layers of hackery to try to find it and suppress it. It seemed at first that every additional intervention attempted just seems to add more whitespace, so I was really struggling for a while, but came up with a strategy to preserve the required whitespace early in the process, then aggressively prune the final result. I think that’s working now, pending more proofing from SK and EH.
On Thursday, went through the rest of the TODO document, and fixed a few things; others were more along the lines of ideas for enhancements and questions, which I responded to. There are a couple of font-level infelicities which we might end up trying to work around by means of a final transformation of the XSL:FO itself, making subtle adjustments to font sizes, but it remains to be seen whether that’s completely necessary.
On Friday, met with SK and discussed these changes and the feature requests.