G’day Comrades,
Continuing to have issues with the Vietnamese parser (raised here, but it seems to be deteriorating).
Per screenshots attached, text copied in to the input feature is not accurately replicated in the actual lesson. I imagine this is some sort of attempt at text normalisation, but (as raised in the link above several months ago which does not seem to have been actioned in any way) the parser doesn’t seem to be appropriately handling the different ways Vietnamese text can be encoded (precomposed as a whole character complete with diacritics, or decomposed as a base letter with additional encoding for diacritics).
As a result, some text (probably decomposed) is being rendered incorrectly - see example below for perfectly legible input text being presented with missing and duplicated letters. As suggested previously, this (as well as the duplicate LingQs issue previously complained of with no action) would likely be fixed with better text normalisation that just put all vietnamese input into precomposed (NFC) form. Vietnamese is not a beta language - it needs to be able handle this.
This isn’t an open source project - it’s paid (quite well at the Premium Plus tier). The recent issues on misaligned LingQs occurred in the several days leading up to a certification exam for me, causing quite a bit of disruption to something quite important. It is becoming harder and harder to justify paying for this service when these issues (which impact basic features of the site) occur regularly and without warning, and only seem to be actioned after widespread complaints.
Would greatly appreciate some info on the plan to remediate this and the other duplicate LingQ issue that seem to both be caused by text normalisation.
