Greedy printing?

[jennybc]

I run into this fairly often: a character vector that has some long entries and makes the printed tbl_df very disappointing. Since it can't be printed in its entirety, nothing gets printed at all and leads to a huge amount of sad whitespace.

Could printing just use all the available horizontal space and truncate the last variable? I realize it's hard to say whether the first part or the last part is more interesting to reveal. And that this feature is probably a pain in the rear to implement.

## the type variable is completely left out :(
xlsx_read_workbook_rels(ek_path)
#> Source: local data frame [3 x 3]
#> 
#>                     target                id
#>                      <chr>             <chr>
#> 1    /xl/sharedStrings.xml Rfd88d28f71e84f97
#> 2           /xl/styles.xml Ra71ceb88d7f8404d
#> 3 /xl/worksheets/sheet.xml R6082ddd3e995440f
#> Variables not shown: type <chr>.

print(xlsx_read_workbook_rels(ek_path), width = 300)
#> Source: local data frame [3 x 3]
#> 
#>                     target                id
#>                      <chr>             <chr>
#> 1    /xl/sharedStrings.xml Rfd88d28f71e84f97
#> 2           /xl/styles.xml Ra71ceb88d7f8404d
#> 3 /xl/worksheets/sheet.xml R6082ddd3e995440f
#>                                                                                type
#>                                                                               <chr>
#> 1 http://schemas.openxmlformats.org/officeDocument/2006/relationships/sharedStrings
#> 2        http://schemas.openxmlformats.org/officeDocument/2006/relationships/styles
#> 3     http://schemas.openxmlformats.org/officeDocument/2006/relationships/worksheet

## would be nice
xlsx_read_workbook_rels(ek_path)
#>                     target                id                   type
#> 1    /xl/sharedStrings.xml Rfd88d28f71e84f97 ...ships/sharedStrings
#> 2           /xl/styles.xml Ra71ceb88d7f8404d ...elationships/styles
#> 3 /xl/worksheets/sheet.xml R6082ddd3e995440f ...tionships/worksheet

[krlmlr]

Why only the last column? We could look at evenly chopping space off each column by e.g. reducing precision or truncating. See #51 for pending formatting updates.

[jennybc]

That had crossed my mind, but I thought it would be a nonstarter. It is interesting.

But also this: I usually put specific identifying variables up front and I would hate to have less information displayed for those in order to fit in truncated versions of other variables I am less interested in.

I could go either way. It just feels like the big swath of whitespace is a lost opportunity to print something about the variables that don't completely fit.

[krlmlr]

Extra whitespace is also often caused by column names that are too long. Maybe truncate them to the length of the column, and show the full name below if necessary?

We could keep the full width for the first 2 or 3 or so columns. Anyway, I guess showing a part of the last column is not that difficult, we could start with that.

[jennybc]

Yes, it feels like column names should never be allowed to hog horizontal space and force other variables off the edge. Truncating them to the width required by their data seems sensible.

[krlmlr]

• add ... in the middle of the string
• general threshold for string columns: 15 or so
• column names are truncated if total width is too large

[hadley]

Related, I wonder if tibble should print with a smaller number of digits that getOption("digits") - the default of 7 is really too high, and IMO 3 or 4 would be much better. This is somewhat tangled up with #30 as digits doesn't really control the number of significant digits.

[jankatins]

I'Ve added a first implementation of the "shorter string" idea in #104

[krlmlr]

We've implemented adaptive widths in colformat, which is used as of now. Can you please try, and report any problems to the colformat repo?

[github-actions]

This old thread has been automatically locked. If you think you have found something related to this, please open a new issue and link to this old issue if necessary.