Excel data table does not work in Gnumeric
Submitted by hayden
Link to original bug (#722059)
Description
I posted the below to the gnumeric list and was told to file a bug report here.
Short Version: I created a data table via the menus in Excel 2000. It works fine when I open it with Excel 2000 on another machine, but does not work in Gnumeric (or Kingsoft or OpenOffice).
Details I am running a simulation and gathering the results in a data table. The table should have a result for each time the simulation is run (just 25 times for testing purposes). If I press F9, I should get another, different set of 25 results. This works in Excel 2000. Ultimately it is intended for educational use and I would like it to run in a free spreadsheet. Opening the file in gnumeric 1.10-16 it looks fine but when I press F9 I just get 25 copies of the last run instead of results for 25 different runs. Using Kingsoft 2012, the spreadsheet looks OK on opening but if I press F9 I just see the original 25 outcomes. This is all on the machine where I created the file, and that version of Excel has several add-ins, so I copied the file to another machine and did a clean install of Office 2000. My spreadsheet still works fine in Excel 2000 without add-ins. This machine has OpenOffice 3.4 on it and when I open the spreadsheet there it is already 25 copies of the same number instead of 25 different outcomes. I installed gnumeric 1.12-9 on the second machine and it is just like 1.10-16 (does not work). It works in Excel 2000 in CrossoverOffice on a third machine running Linux. From my vantage point of total ignorance I am wondering if the alternate spreadsheets decided to speed things up by not recalculating the table "if nothing has changed". I admit the table is odd in that the input value does not appear in the quantity being calculated. In Excel, changing the input value prompts a recalculation anyway, and the result is different because the output of rand() changes each time you recalculate. As far as I know, what I am doing is standard practice for running simulations in Excel, so it seems to me a bug if gnumeric cannot do this. I am following a paper published 14 years ago. My spreadsheet file is just 20k and contains no macros or VBA -- just stuff from the default menu and function list (AFAIK). I am guessing the first step would be to see if anyone else can get this to run in Excel and confirm that it does NOT run in gnumeric. The file is at http://statland.org/temp/permtest3.xls. Cells G5:G29 is where the different programs differ.
Version: git master