Let me continue https://samskrtam.ru/textdiff-dictionary/ — comparing of headwords in Sanskrit dictionaries with NLP tools.
Difference between «perfect» devanagari sorting and «usual» Indian sorting is quite huge. Out of 43142 lines 12062 where changed in http://www.sanskrit-lexicon.uni-koeln.de/scans/SHSScan/2014/web/index.php How can one find a word if one does not where to look for 1/3 of entries?