@oscargus can you check this pull request that doesn't have conflicts?

This looks fine from that aspect. Still, there are some linting errors.

This looks fine from that aspect. Still, there are some linting errors.
Could you advise me on how to solve them?

You can view the linting errors by clicking the “files changed” tab at the top of this page and scrolling down to see the comments from “reviewdog”.

All checks are passed! Can we merge it because I have deadline till today for a university project and I will enrich in the next days to be more efficient?

though I'd prefer a comment on how it's making decisions.

Edit: I cannot even get my head around what it does. But now I think I am correct after several attempts. If basically takes the smallest number between 4 and 9 such that the number is not a integer factor of the number of elements.

max_reminder is never updated and the code can be rewritten as:

def columns_calculation(n):
    max_columns = columns = 10
    for columns_number in range(max_columns - 1, 3, -1):
        if n % columns_number:  # Not integer multiple
            columns = columns_number

    return columns
    
nvals = range(1, 100)
cols = [columns_calculation(n) for n in nvals]

import matplotlib.pyplot as plt
plt.bar(nvals, cols)

So putting 4 is also a quite good heuristic.

You may want to check the discussion in #26183 @story645

My standpoint is still that this is not what is wanted.

@oscargus I think you're right that a better solution would take word and glyph length into account - there are still some lines that are not great-and use one or more of the dynamic reflow sphinx options.

I also think this solution produces easier to read tables than what is already there & removes the need to manually specify columns and is therefore a good incremental solution.

Better incremental solution that gives easier to read tables: 4. (I now got my head around the logic, so please reread the edited post.)

The only table that has 5 is because there are 32 items in it. Adding one and it will be 4 as all the others.

Changing the number of items, as is done in #26173 and #26176 will modify the outcome (although I have not bothered to compute the number of entries there).

I'd say that picking a column number based on a probability as 4: 75%, 5: 20%, 6: 4%, 7: 1% is a comparable approach.

Ok! I will look at it! Thank you!

I was thinking about finding the sum of characters on each line and choose how many columns (from 4 to 7) will have the most equally distributed summary on each line.

I'd suggest finding the longest string in the list/set in syms and then use that to figure out a suitable number of columns.

If you're gonna do this, you're gonna want to probably make the (latex, symbol) tuples and use that as longest symbol, and really this becomes an optimization problem & at that point it's likely easier to see if it's already solved by pretending it's a table of individual elements and using auto https://sphinx-design.readthedocs.io/en/latest/grids.html#item-level-column-width

ETA: Honestly I also think @oscargus's set width to 4 solution - where you strip the function call out and replace it w/ 4 is an improvement.

@NikosNikolaidis02 are you still interested in working on this?

I am going to close this one as it has stalled for a while and comments above state that the current implementation is not what is wanted. @NikosNikolaidis02 please comment if you are still interested in working on this and we can re-open.