print col.names='none' does not affect column width

In print.data.table when setting col.names='none' I expected column widths to be sized according to their contents. Instead, the column widths are unchanged. Incidentally, when class is suppressed the column widths are different. I expect the behavior to be consistent between those two options.

library(data.table)
dt<-data.table(x=3:1,sillylongcolumnname=1:3)
print(dt)
#>        x sillylongcolumnname
#>    <int>               <int>
#> 1:     3                   1
#> 2:     2                   2
#> 3:     1                   3

print(dt,col.names='none')
#> Warning: Column classes will be suppressed when col.names is 'none'
#> 1: 3                   1
#> 2: 2                   2
#> 3: 1                   3

^{Created on 2025-03-24 with reprex v2.1.0}

@tdhock

Hi @tdhock ,
I'd like to work on this issue — could you please assign this issue to me?

Here’s my proposed solution:
I plan to modify the width calculation logic so that:

Column widths are based only on actual data content when col.names = "none".
Width contributions from column names and class annotations are skipped in that case.
Behavior remains unchanged when column names are printed normally.

This should result in a cleaner and more compact output when column names are omitted.

Let me know if this approach works, and I’ll move ahead with the implementation. Thanks!

@venom1204

Yes, go ahead @venom1204, that's the basic idea.

@aitap

hi @aitap ,
After further investigation, I found that the most robust solution is to explicitly set all column names to empty strings in the toprint object when col.names = "none":

if (col.names == "none") 
  colnames(toprint) <- rep.int("", ncol(toprint))

I plan to add this line in the print.data.table function, right after toprint is constructed (e.g., after
toprint = x[idx, ] or toprint = x), but before any formatting or width calculations. This ensures all downstream formatting uses empty column names, so column widths are determined solely by the data, fully resolving the issue.

All tests, including edge cases, now pass and the output is compact and consistent.
Could you confirm if this approach is acceptable? Also, if there are any potential side effects or unrelated areas of the codebase that might be impacted by this change(right now there aren't any according to me)—especially ones I may not be aware of—please let me know. If approved, I’ll proceed with the implementation.

thank you.

Indeed that sounds good. Potentially it could just be included as a condition here:

data.table/R/print.data.table.R

Lines 94 to 96 in 2b7eb4c

    
           if (is.null(names(x)) || !any(nzchar(names(x), keepNA=TRUE))) 
        
             # fixes bug #97 and #545 
        
             colnames(toprint)=rep("", ncol(toprint))

(textproc/R-data.table) Updated 1.16.4 to 1.18.4, Fix build against 4…

Navigation Menu

Repository navigation

print col.names='none' does not affect column width #6882

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

	if (is.null(names(x)) \|\| !any(nzchar(names(x), keepNA=TRUE)))
	# fixes bug #97 and #545
	colnames(toprint)=rep("", ncol(toprint))

print col.names='none' does not affect column width #6882

Description

Activity

venom1204 commented on Apr 9, 2025

MichaelChirico commented on Apr 9, 2025

venom1204 commented on Apr 17, 2025

MichaelChirico commented on Apr 17, 2025

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions