Search behavior for mixed-case item names
Description:
Today, mixed-case names appear to be split at uppercase→lowercase boundaries when indexed. That helps for classic camelCase, but it breaks acronyms when the casing is slightly different.
Examples of how names seem to be tokenized
XMLParser -> XML + Parser
XMLparser -> XM + Lparser
Problem:
For both forms, searching "XML" should return the item.
It works for XMLParser, but not for XMLparser, because the acronym is not kept as a single token in the second case.
We see other patterns, such as:
TOPaaa → searchable with TO or TOPa, but not TOP
WORKaaa → searchable with WOR or WORKa, but not WORK
Request:
When an uppercase run looks like a meaningful word (e.g. XML, TOP, WORK), please index that full uppercase segment as its own token—even when it is immediately followed by lowercase letters (XMLparser, TOPaaa).
Thanks,
Anonymous
shared this idea