14.4 Bug Update

Problem with TEXTJOIN and “do not include empty cells”. FIXED!!!

TEXTJOIN would include cells that were null strings. You might end up with extraneous commas in the result. This made us jump through extra hoops to remove those extra commas. It now ignores null strings the same as actual empty cells.


The bug was present in 13.1 but is fixed now.


Wrapped text in a popup may cause tables to overlap. FIXED

If a table has another table below it and a cell increases in height in the upper table, the lower table is supposed to move down accordingly to accommodate the increased height.  This worked fine except if that wrapped cell was a pop up cell.  The cell height would increase but the table below would not move down and might end up getting overlapped.  

----

Sorry to say that that is the extent of the bug fixes for bugs I have reported. Now onto the ones that did not get fixed.


Bug related to cell borders not displaying or printing correctly.  Bug remains in version 14.4.

Cell borders that are the same color as the cell fill will be visible if they are less than 1pt width. They appear to be a lighter shade. If set to 1pt or greater the problem goes away.


Tables with no cell borders print with faint borders. It looks correct in Numbers but not when printed.


Bug was first reported in version 11.2 in October 2021. Still remains in version 14.4 in April 2025.  Three and a half years and counting….


Calculation Engine Bug. Bug remains in version 14.4

Changes Apple made to the calculation engine deviated from the IEEE 754 standard and have rendered Numbers useless for recursive formulas that require high precision. 


Numbers used to follow the IEEE 754 standard religiously. This caused some “user expectation” problems when numbers did not add and subtract exactly like they do on paper, sometimes resulting in a very small error instead of being exactly zero. Apple apparently decided to massage the results to “fix” this.  But in doing so they created other small errors, or at least inconsistencies with other math apps such as Excel and Matlab.


Bug was first reported in version 12.0 in March 2022. I believe it was around many versions earlier but I do not recall what version had the change to the engine and did not find it then. Still remains in version 14.4 in April 2025. Far more than 3 years and counting…


Reference Errors When Sorting. Bug remains in version 14.4

This is a very glaring bug and it really irks me that is is still not fixed. Formulas such as SUM and COUNTIF when used with a range of the form $A2::A2 will develop reference errors if the table is sorted.  Example is COUNTIF(A$2:A2,A2) filled down to the rest of the column for the purpose of finding duplicates in column A (rows where the formula result will be >1). This is an extremely common referencing method for formulas that look only at the top of the table (from the current row and up) and it used to work correctly, even after a sort.


Bug was first reported in version 10.0 in June 2020. It still remains in version 14.4 in April 2025.  Almost five years, yes FIVE YEARS, and still counting…


INDIRECT, OFFSET, INDEX, and FORMULATEXT do not work correctly when referencing cells in categorized tables.  Bug remains in version 14.4

When changes are made to a categorized table and rows move around (such as sorting), these formulas do not keep up with the changes. It is as if they are still referencing the prior version of the table. For example, say INDIRECT(“Table 1::B5”) is giving the result “4” (the value in cell B5). After a sort, that row moves somewhere else and now cell B5 has “10” in it.  The formula still has “Table 1::B5” but the result is still “4”, not “10”.


iMac 27″ 5K

Posted on Apr 14, 2025 2:33 PM

Reply
4 replies
Sort By: 

Apr 14, 2025 2:34 PM in response to Badunit

Print Rendering Causes Overlapping Tables. Bug still remains in version 14.4

If a table is longer than one page and if there is another table located below it, the first table will end up overlapping the one below when printed. It is much worse if the first table has repeated headers on each page but it happens either way. 


Bug first reported in version 13.0 in April 2023. Bug still remains in version 14.4 in April 2025.  Two years and counting…


Bug in Value Labels on Scatter Charts. Still present in version 14.4

When “Show Series Name” is selected for a series, the chart displays the series name at all X-Y pair data (as it should) but also puts the series name for the first row of the range and the last row of the range even if those cells (the Y values) are blank in the table. No other blank rows get a label, just the first and last.


To "highlight" a specific data point, its Y value can be moved to a new series all by itself. Using "show series name" allows you to put a label directly on that point. But now it also puts a label on two other "points" that do not exist. That ruins the idea.


First reported in version 11.2, probably in 2021 I think but it may have been around before. Still remains in version 14.4 in April 2025. Over three years and counting…


Bug using dates in SUMIFS (and probably other functions) when changing regions. Still remains in version 14.4

If a date is a condition in SUMIFS, the result will not be correct if the region is changed from US to Canada.  It appears that if the US date format is not one of the built-in ones for Canada, SUMIFS will misinterpret it. The MONTH, DAY, and YEAR functions get it right but SUMIFS does not. The date in the example I reported is in a cell formatted as date&time, it is not hard coded text in the SUMIFS.  This probably applies to other functions and other region changes as well.


First reported in version 11.2, probably in 2021. Still remains in 14.4 in April 2025. Over three years and counting…


Trendline Equation Bug

In some instances the trend line equation on a chart will be “y=0” when this is obviously not correct.


First reported in version 12.2.1. Still remains in version 14.4.  How many years and counting?


WORKDAY function gives incorrect results when the date is the same as a skip day.

For example, if the date is Tuesday and that date is one of the skip days and the number of workdays to “skip” is -1, the first valid workday one day back is Monday but it will skip back two workdays to the previous Friday. Similar with +1, it skips Wednesday and gives you Thursday.  It can be argued that the function is undefined for this particular example but the results seem incorrect to me.


This bug has been around since at least version 12.0.


New (?) bug that I’m not even going to bother reporting:

Numbers is showing a mirrored image of a textbox. This persists after closing and reopening the document.



Reply

Apr 14, 2025 7:59 PM in response to Badunit

Thank you! On the sorting bug it SEEMS as if the new SORT and SORTBY functions will help as a workaround. And the old ways of finding duplicates or extracting distinct values are close to extinct with UNIQUE. But they should still squash the bugs!


SG

Reply

Apr 15, 2025 4:44 PM in response to Camelot

Camelot wrote:
I don't understand this one. Isn't that just the text box reflection?


Sure enough, that's what it was. Never clicked that before, don't know how it got clicked. Problem solved.


If you can solve a few of the others too I'd really appreciate it.

Reply

14.4 Bug Update

Welcome to Apple Support Community
A forum where Apple customers help each other with their products. Get started with your Apple Account.