adjust column width with putdocx

[Simon Pfaff]

Hallo,

I've got a question concerning the column width of tables and its assignment with "collect style putdocx". I understand that the sum must be equal to 100. However, even if the sum does equal 100, in some cases I get error messages implying that this is not the case.

Are there any rules to be considered, when defining column width?

clear all
set more off

sysuse auto.dta

putdocx begin

table foreign rep78, statistic(freq) statistic(sumw) statistic(percent, across(foreign)) ///
                      sformat("%s%%" percent) nformat(%9.0f) totals(foreign)

* does work
*mat width1  = (37, 15.5, 9.5, 9.5, 9.5, 9.5, 9.5)
*mat width1  = (37, 15, 9.6, 9.6, 9.6, 9.7, 9.5)
 
* does not work
mat width1 = (37, 15, 9.6, 9.6, 9.6, 9.6, 9.6)
 
collect style putdocx, halign(left) width(80%) width(width1)

putdocx collect

putdocx save test, replace

. putdocx collect
option width() specified incorrectly
    The sum of the column widths specified in matrix width1 must be equal to 100.

What do I do: I generate a series of tables with table/collect layout and arrange them one below the other. The tables may have a different number of columns. However, the first column (with labels) and the second columns should have the same width. The other colums may vary in width.

I define several vectors "width#" to set the width of each table. However, sometimes it is pretty difficult to find a vector which is working. The slightest change will result in the procedure either working or not. I don't really understand why this is the case.

Thanks a lot!

[Zhao Xu (StataCorp)]

Simon Pfaff,

Thank you for bringing this to our attention. This is a bug caused by the precision issue. The sum of the vector that does not work, i.e., (37, 15, 9.6, 9.6, 9.6, 9.6, 9.6), does not have an exact finite-digit binary representation, compared with the other two vectors.

We will fix the width() problem in putdocx to make it more robust to the above cases and release it to users in one of our future updates.

[Simon Pfaff]

@Zhao Xu (StataCorp)

Thank you for your feedback, I appreciate it!

Best, Simon