search

atom / fuzzaldrin

Fuzzy filtering and string scoring
MIT License
319 stars 28 forks source link

Rewrite scoring algorithm to support run of consecutive character, fix acronyms and add optimal selection of character. #22

Closed jeancroy closed 9 years ago

jeancroy commented 9 years ago

This will be available for testing as an option in fuzzy-finder

Screen

Please address problems, comments, and suggestions here: https://github.com/jeancroy/fuzzaldrin-plus

We start a tuning phase, so anything not quite right is welcome.

benogle commented 9 years ago

Yeah, I suppose you will need to override populateList

jeancroy commented 9 years ago

As requested there is now a PR in fuzzy-finder.

benogle commented 9 years ago

This has been added to atom/master via https://github.com/atom/fuzzy-finder/pull/142. If you build atom from source, it is available via option:

Screen

benogle commented 9 years ago

Closing this as we can continue the discussion over at https://github.com/jeancroy/fuzzaldrin-plus

jesseleite commented 9 years ago

Thank you <3

jeancroy commented 9 years ago

For those who want to test it, fuzzaldrin-plus will be an option of autocomplete-plus 2.22

jesseleite commented 9 years ago

Awesome, look forward to it!

AbdullahAli commented 8 years ago

The alternate scoring option makes the search really slow and "jittery" :-1:

jeancroy commented 8 years ago

Hi @AbdullahAli , can you give an example of query and context ? Is it fuzzy finder, how many file indexed ?

The benchmak show it'll run anywhere between 2x faster to 2x slower depending on query. (And for most of the query, on most condition I expect to err on the side of faster.)

Are you using this from a official stable distribution of atom or using some hack needed before ?

jesseleite commented 8 years ago

I haven't found any performance issues with alternate scoring option. It's been great for me! @jeancroy how do we find out how many files are indexed?

jeancroy commented 8 years ago

Very first run of fuzzy finder in a project. If you have many file it show a waiting message with number of file so far. img

But to create such scenario I have to open atom in my document folder or something multi-project like that.

jesseleite commented 8 years ago

No way to tell index count after it's already indexed? Maybe via console command or something?

AbdullahAli commented 8 years ago

Hey @jeancroy I have ~36k files

screen shot 2015-11-18 at 16 37 31

This is what it looks like when the alternate scoring option is enabled: https://vid.me/NBz0

This is what it looks like with it disabled: https://vid.me/hQUp

In the second video you can see it is smoother and not "jittery" - although the video quality might not be the best, so sorry about that

jeancroy commented 8 years ago

In both video it show "reindexing project" so that does not helps. What you see as "jerky" is that you type faster than the time needed to compute a result.

I'll open an issue on fuzzy finder about the debounce delay. (Maybe making a search each 2 char will help reduce pressure)

Btw what kind of system do you have ? There's a lot of ruby user that use this and the difference should not be that large.

AbdullahAli commented 8 years ago

It is the same effect after the "reindexing project" message goes away.

I am using a macbook pro on yosemite ![Uploading Screen Shot 2015-11-18 at 17.19.46.png…]()

jeancroy commented 8 years ago

Ok so a recent build? Are the file-name confidential ? Would you mind sharing the file listing ? Maybe to a gist.

You can save it to a file like so:

  find /Path/To/Project -type f > FilesWithPaths.txt

I'm ok with you redacting out name of company , software, and whatever it need to be comfortable. Ideally replacing by another string of the same length.