gdh1995/vimium-c

Omnibar searches rarely filter to the expected item

Opened this issue · 1 comments

shallowbit commented

DESCRIPTION:

his one is hard to pin down and describe. Although, I'll attempt to. the scenario below is invented but is so merely to illustrate the issue.

When opening the Vomnibar (VB) to "tabs", you are shown a number of tabs and their respective URL's.

Say these tabs are shown:

cnn.com - the place for news
  cnn.com/frontpage
craigslist - missed connections
  cragslist.com/mc
Corgis are the best : "can i haz cheezeberger"

searching for 'co' will select some random URL without "Co" anywhere in the tile because ".com" is in most URL's (at least for me). Also, searching for "cnn" will many times not select the top result in the Vomnibar but show another unexpected result first.

SUGGESTION:

I think most people usually search by title most of the time. I think weighting results more on what's actually shown in the Vomnibar when triggered would help, at least in certain scenarios like when activation only show's tabs or bookmarks. "Vomnibar:history" could possibly weight the result's evenly? regardless, I find myself almost always having to be overly verbose to find things by searching and I think the system as whole could use revisiting.

Possibly some options like:
Vomnibar:VisablePriority
VomnibarTabSearch:TitlePriority
Vomnibar:IgnoreURLS

I think this is fairly universal, but I'll include version info. I hope this illustrates the issue.

ENVIRONMENT:

  • Browser name: Edge
  • Browser version: 123.0.2420.81 (Official)
  • Vimium C version: 1.99.99
  • OS name and version: Windows 11
gdh1995 commented

Um sorry but Vomnibar indeed encourages users to type long words. Now that it sorts matched history items by both word similarity and visiting time, if you just visited a page a few minutes ago, then Vomnibar should give it a much high score.

co does conflict with .com frequently, but this is just a special case - most English words don't occur in web hosts.

The logic to calculate final scores is very hard to upgrade - I have no way to recognize whether a new logic is better or worse.