Make default column width for Data Explorer tighter (sorting indicator space)

[jennybc]

This is sort of a resurrection of #6943, although our current situation is not as wide as things were when #6943 was first opened.

But I generally would like support for making columns fit their data. I see that #3259 tracks the gesture for shrinking 1 column to fit its data.

But I'd love the data explorer to just open to all the columns "hugging their data". Or having a button I could click to do that.

Here's a little screen recording of me showing that the default view opens with many columns wider than their data. It's frustrating to watch, because I have a hard time grabbing that "handle" but that is exactly why I would love to not do this manually for individual columns:

resizing-columns.mp4

System details:

Positron and OS details:

Positron Version: 2025.10.0 (system setup) build 170
Code - OSS Version: 1.103.0
Commit: bfddb69
Date: 2025-10-01T07:33:41.572Z
Electron: 37.2.3
Chromium: 138.0.7204.100
Node.js: 22.17.0
V8: 13.8.500258-electron.0
OS: Windows_NT x64 10.0.26100

Session details:

Describe the issue:

The typical column seems wider than it needs to be (where "need" is defined by the width of its name and data, plus the vertical dots we add). But to make everything "tight", the user has to drag something fiddly, column by column.

Steps to reproduce the issue:

Get any data frame open in the data explorer. For example, View(penguins) in R.

Expected or desired behavior:

All columns open at the minimum width before the header or data gets truncated and displays ....

Or there's a gesture I can explicitly use for this.

Or a setting I can configure to opt in to this.

Were there any error messages in the UI, Output panel, or Developer Tools console?

[juliasilge]

Right now, the default width gives space for the sorting indicator:

[jennybc]

As an infrequent user of sorting, I think I experience that as unused horizontal space. I'm optimizing for "number of columns I can see".

[jennybc]

Just thinking about the prevalence, so to speak of sorted columns ... I think it's helpful to think about what proportion of the time a user has any sorted columns. And, when they do, how many columns are likely in a sorted state. Practically, the number of sorts you can have active is pretty small. So I wish the default width didn't reserve real estate for the sorting indicator. Maybe a sorted column should be viewed as the exception, rather than the rule?

[anbrav0]

Thank you for having this issues available @juliasilge and @jennybc ! I am kindly adding some comments about this from someone who has limited screen space (smaller display) when, for example, displaying a table in the editor, it feels like the column width for the data explorer could be a bit tighter for some columns, as well as the row column. I have attached a video demoing as an example.

var-column-with-could-be-tighter.mov

in the video you'll notice that column house_style shrinks a little after sorting, which is a little bit of a nicer display. it's not included in the video but removing of the sorting keeps that tighter column width.

I've also included a before and after of manual resizing of the columns for some sort of "ideal" column width size.

before

after

R code for repro

Positron Version: 2026.04.0 build 172
Code - OSS Version: 1.109.2
Commit: 16244258acc309ddd8e1b53a91bc27ed293b95c5
Date: 2026-03-18T16:16:42.295Z
Electron: 39.3.0
Chromium: 142.0.7444.265
Node.js: 22.21.1
V8: 14.2.231.22-electron.0
OS: Darwin arm64 24.6.0

# Load libraries needed to create viz requested from book: -----
# https://r02pro.github.io/global-local-mapping.html

library(r02pro)
library(tidyverse)

# Question 2: In the plot from Q1, using different colors for both the
# scatterplot and the smoothline fit to distinguish whether the house has more
#than 3 bedrooms and make all points to be size 2. ----

# create new binary variable
sahp_new_df <- sahp |>
  mutate(
    bedroom_binary = case_when(
      # create binary var <= 3 / > 3 bedrooms
      bedroom %in% c(0, 1, 2, 3) ~ "<= 3 bedrooms",
      bedroom %in% c(4, 5) ~ "> 3 bedrooms",
      .default = NA_character_
    )
  ) |>
  relocate(bedroom_binary, .after = bedroom)

# 1. add scatter plot color for new binary bedroom bar 2 make the smooth line
# size 2. I _think_ thats what the question wants? ----

ggplot(data = sahp_new_df) +
  geom_point(
    aes(x = liv_area, y = sale_price, color = bedroom_binary),
    size = 2,
    alpha = 0.5
  ) +
  geom_smooth(
    aes(x = liv_area, y = sale_price),
    method = "lm",
    color = "lightgreen"
  ) +
  theme_bw(base_size = 11) +
  theme(
    legend.position = "bottom"
  ) +
  # add labs
  labs(
    title = "Living Area and Sales prices by \n # of bedrooms"
  )

# create plot to visualize some outliers ----

ggplot(data = sahp_new_df) +
  aes(x = liv_area, y = house_style, color = house_style) +
  geom_boxplot() +
  # add random theme
  theme_bw(base_size = 11) +
  theme(
    legend.position = "none"
  )

# show dist of sales prices ----
ggplot(data = sahp_new_df) +
  aes(x = sale_price) +
  geom_histogram() +
  # add labs
  labs(
    title = "Distribution of Sale Prices"
  )

# show dist of liv_area -----
ggplot(data = sahp_new_df) +
  aes(x = liv_area) +
  geom_histogram() +
  # add labs
  labs(
    title = "Distribution of Living Area"
  ) +
  # adding a random theme
  theme_bw(base_size = 11)


# briefly summarize ----
sahp_new_df |>
  summarise(
    mean_live_area = mean(liv_area, na.rm = TRUE),
    mean_sale_price = mean(sale_price, na.rm = TRUE),
    median_sale_price = median(sale_price, na.rm = TRUE),
    median_liv_area = median(liv_area, na.rm = TRUE),
    .by = house_style
  ) |>
  arrange(desc(median_sale_price)) |>
  view()

[codeswitching]

This may be a separate issue/bug, but for me, the column spacing is sometimes much larger than what is needed for the sorting indicator. Here are some examples taken right after opening a new data explorer instance on two different data frames: