Searching depends on the order of the searched words

[n7800]

There is a problem with the word order in the search. Perhaps it does not concern you, but GitLab itself:
https://gitlab.com/gitlab-org/gitlab/-/issues/414997#note_1426833558

[jeancroy]

This is by design. Inherited from the fuzz-aldrin library that we tried to replace. That was the behavior in atom text editor and other similar programing text editor (notably sublime) Basically the query must be a subsequence of the text that is searched. Then we do mathematical optimization to group those into large subsequences of consecutive characters. Jean Christophe Roy

…

On Thu, Jul 13, 2023 at 12:47 PM n7800 ***@***.***> wrote: There is a problem with the word order in the search. Perhaps it does not concern you, but GitLab itself: https://gitlab.com/gitlab-org/gitlab/-/issues/414997#note_1426833558 — Reply to this email directly, view it on GitHub <#42>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADBMEES2ILJUZSVZ4HEIR43XQARCTANCNFSM6AAAAAA2JG25MI> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[jeancroy]

It would not be very hard to split query into tokens (say by spaces)
Then score every entries against each tokens.
Then combine those in an appropriate way. (we can start with a sum of scores)

But experimentally path contain a lot of words and typing subsequences is often the fastest way to remove non interesting entries fast. It just requires some time to get used to it. Maybe you can add query as a whole and each score individually to get a compromise.