The current system wrongly indexes entries containing non-alphanumeric characters. For example, you cannot put an apostrophe at the beginning of an entry because this causes it to be indexed under the apostrophe, rather than the letter! And if a non-alphanumeric occurs in the middle of an entry, the entry will be listed in an unexpected order. LO needs to implement an accepted worldwide standard which allows for non-alphanumeric characters: i.e. the LETTER BY LETTER system of alphabetising indexes. This system is detailed at https://www.press.uchicago.edu/Misc/Chicago/CHIIndexingComplete.pdf (Chicago Manual of Style. To give an example. This (AFAIK) is a CORRECTLY indexed series of entries (letter by letter): "Bell, Andrew, 413 Bell, book, and candle 36 Bell, Garden, the 175, 651 Bell Hotel, The 559pp. Bellmen 116 Bell-ringers, the 205, 380, 578 Bells of the Parish Church 173" However, this is how it is INCORRECTLY indexed by LO: "Bell Hotel, The 559pp. Bell-ringers, the 205, 380, 578 Bell, Andrew, 413 Bell, book, and candle 36 Bell, Garden, the 175, 651 Bellmen 116 Bells of the Parish Church 173"
As far as I can see, this reports leads to the general and very important question, if LO should follow the rules for a special style (for example Chicago Manual of Style). So I would change the bug summary to "Alphabetical index should follow rules of Chicago Manual of Styles". Decision about that is not trivial. Perhaps alphabetical index follows the rules of a certain style guide. If this is the case, there should be an information in LO help [1] cc: Design Team for further dicussion an decision [1] https://help.libreoffice.org/7.0/en-GB/text/swriter/01/04120212.html?&DbPAR=WRITER&System=WIN
Guess the sorting algorithm would be the same for Writer and Calc but it isn't. Trying with 6.4 it returns "Bell, Andrew, 413 Bell Hotel, The 559pp. Bell-ringers, the 205, 380, 578 Bell, book, and candle 36 Bell, Garden, the 175, 651 Bellmen 116 Bells of the Parish Church 173" in case of Calc but "Bell... at the end in case of Writer, alphabetical ToC.
UX-wise the sorting would go into the "key type" dropdown as an alternative to alphanumeric. So introducing another algorithm wouldnt be a problem at all and is welcome.
*** Bug 129500 has been marked as a duplicate of this bug. ***
Another aspect of this problem: the Index does not allow you to have entries under the SAME alphabetical delimiter which have the ame letter but different diacritical marks. So, for example, A, Ā, À, Á etc. are placed under DIFFERENT alphabetical delimiters! There's a section about alphabetization of diacritical characters on Wikipedia: https://en.wikipedia.org/wiki/Diacritic#Alphabetization_or_collation
See also, Bug 140186: Index needs to ignore diacritics when alphabetising.