[Libreoffice-ux-advise] [Bug 149984] Add sortable counts to AutoFilter drop down

[bugzilla-daemon at bugs.documentfoundation.org]

https://bugs.documentfoundation.org/show_bug.cgi?id=149984

--- Comment #6 from Pierre Fortin <pf at pfortin.com> ---
Created attachment 181265
  --> https://bugs.documentfoundation.org/attachment.cgi?id=181265&action=edit
Mock up of suggestion

(In reply to Heiko Tietze from comment #5)
> Please have a look at the Data Provider feature. Being kind of experimental
> it's not yet included in the UI, you have to add the commands via Tools >
> Customize. The DP allows you to filter and manipulate data before loading
> into the sheet - and if you connect to the same source it's just an update
> later.

Sounds interesting; we may be able to use this for final reports; asuming this
can be used to conditionally load data. It would not help during our
investigation phases, we need all data where we do whatifs based on text
content. Much easier and faster than running a series of SQL queries, where
modifying the query, rerun, view new results would slow down the "didn't expect
xyz, filter out a,b,d,g,x from another column; oh wow!" process.

> Summary from the last GSoC project is here:
> https://libreoffice-dataproviders.blogspot.com/2021/08/data-providers.html
> 
> 
> Back to the actual request: Enhancing the sort function with count sounds
> like a can of worms to me. Why not Sum()? 

This data has nothing to Sum; it's all about counts of content.
CountIf('Active') would make more sense; but we are often surprised by new
values... AutoFilter is a much faster process for us.  I work with a team of
individuals, all from home.  [BTW, one member reports that LO 7.5 (Windows)
crashes on save (any format); otherwise, he's happy with his first time loading
8M+ rows, formatting and editing.  Doubt he's the type to file bug reports
though, and I don't have his details.]

> And UX wise it adds quite some
> complexity where average user wrap their mind around what "Sort Ascending by
> Count" could mean.

Agreed. BTW, what does "Filter by Condition>Top 10" really mean?  :)  It sounds
like this may be what I want, only worded differently...  
Trying with tiny sheet (240K rows)...
Nope.  Column contains only {Active,Inactive,Temporary} -- Filter by
Condition>Top 10 hides every row. Now, I have no idea what {Top,Bottom} 10
means. Looks like I'll have to dig into the docs...  No time today...

> So my take would be to rather write a macro. Shouldn't be too difficult. If
> the DP is not helpful...

We're dealing with AutoFilter as it is:
- load data
- filter & view results
- next depends on what we see... 
- add/change/remove filters & view new results
- repeat

Writing a macro to replicate AutoFilter (including a UI), with only counts of
data occurrences added...  ;p

Playing with this tiny sheet, I just spotted something totally new which gives
me another issue to dig into...  In this case, it appears there are hundreds of
unique values -- AutoFilter had to discover every unique value and create a
selector for each.  I would think counting each value inside that loop should
be minor. Although the display is more complex... See attachment:  each unique
value has a count, and total of unique values next to All...   HTH

-- 
You are receiving this mail because:
You are on the CC list for the bug.