Table list: sorting breaks when filtered

[bryan-ash]

1. Which driver are you using and version of it (Ex: PostgreSQL 10.0): MySQL 5.7.35

2. Which TablePlus build number are you using (the number on the welcome screen, Ex: build 81): Version 5.6.8 (524)

3. The steps to reproduce this issue:

• Show the Items list
• Observe that tables are sorted alphabetically
• Type to filter the table list
• Observe that tables are sorted by length of name (it appears that underscores are not considered)

No filtering - sort alphabetical

Filtered - sort by name length(ish)!

[huyphams]

Bi @bryan-ash it is not sort by alphabet, it is sorted by fuzzy score, you can change the strategy here

[huyphams]

But I would recommend you to use fuzzy as it will be much faster. For example you can search the table affordable housing hud detail by typing "ahhdd"

[bryan-ash]

@huyphams why does the filter matching strategy change the output sorting order?

[huyphams]

Because the matching strategy is how the app finds and sorts the results, isn't it? Fuzzy matching is the method where the strings are scored fuzzily and then sorted based on that score. So, you can see that matching and sorting cannot be separated.

The "contains" is much more simple, it just filter the string which contains the key word and sort it alphabet.

[bryan-ash]

@huyphams My issue is with how the results of the filter are displayed. This is purely a presentation issue, not related to the filtering mechanism.

• Apple
• Bannana
• Dog
• Orange

Filter this^^^ list for "a" any way you like, what should the list be?

[huyphams]

I don't think you understand how fuzziness works, so it's better to change to the "contains" method, where it should function as you expect.

[huyphams]

The purpose of fuzziness is to prioritize results with higher scores at the top; it is not based on the alphabet but rather on the score.

Therefore, if we filter by fuzziness and sort alphabetically, it contradicts the essence of fuzziness and undermines the convenience that many people expect.

[bryan-ash]

If we're getting personal, I don't think you understand the difference between a filtering algorithm, and a result presentation algorithm.

I totally understand how fuzzy matching works in my editor, I use it all day every day. The results are presented in alphabetical order.

[huyphams]

Let me provide you with a real-life example. For instance, imagine you want to locate a student in a class.

You can find the student by their name and then sort the students based on their height; this involves using the "contains" method. Alternatively, if you're looking for students with specific traits and want to sort them based on the trait score (e.g., math score), you'll notice that students with the highest match scores will be at the top, not necessarily based on their height or any other criteria.

Searching for strings is similar, especially when there are multiple matching methods. "Matching contains" simply searches for a specific pattern, while "fuzzy" searching involves assessing the distance between characters.

Let's consider two names:

a_something_here_t
b_something_there
If the input is "gt," the result "b_something_there" with the g and t closer will be on top. This is the purpose of the fuzzy score system. If you always prioritize the alphabet, you will never get "b_something_there" on top, but instead, "a_something_here_t" will be prioritized.

This is not what people who use fuzzy searching want because they are not searching for terms contained within words. If the words are very long and contain many similar terms, they only want to quickly search by typing some characters, which is much faster.

[huyphams]

[huyphams]

In your list of tables, it's challenging to locate a specific table since they all share the same prefix, "affordable_housing....." However, employing fuzzy matching and entering only "ahb" or "ahc" makes the process much faster, as these are just some characters within the terms.

[bryan-ash]

OK, let's see if I can reframe the issue. The sort order of the Table List is bound to the matching strategy. Please provide a way to toggle the output sorting, separately from the matching strategy.

[chrisk-7777]

I strongly agree with @bryan-ash - this is really confusing behaviour. I actually came to the issue tracker just to report it as a bug and found this issue.

It seems there is some confusion between matching strategy and presentation. The sidebar isn't really a "search" mechanism, its a "filter" mechanism, because by default we already have all available values in front of us. The filter is reducing that set (despite the matching strategy used, fuzzy or otherwise).

From a UX perspective, check out something like the sidebar of: https://developer.mozilla.org/en-US/docs/Web/HTML

Note how all values are presented, but filtering "reduces" the visible set, but maintains the order. Again, its a difference between filtering and search, ordering based on proximity makes more sense when the total set isn't known - like searching the web.