Description: pivot table refresh causes: 1/ no undo 2/ formatting loss Steps to Reproduce: 1. Open the attached "LO63 Calc pivot table formating bugs.ods" 2. Go on the tab "Pivot table BUG and ANNOYANCES" 3. Right-click on cell A1 to select the "refresh". Observe the loss of all formatings 4. On the standard toolbar observe that there is no undo. To verify it, CTRL+Z. Actual Results: 1. formatting loss 2. no undo Expected Results: 1. Refresh ought to keep the formatting of the pivot table. 2. Undo ought to be possible Reproducible: Always User Profile Reset: No OpenGL enabled: Yes Additional Info:
Created attachment 152995 [details] test case Bug also if on XLSX format. Possibly general bug such as 123225 ; 122472 In current LO 6.3 release notes, it is reported that Mike Kaganski, Collabora fixed already some pivot table bugs
Additionnal point of attention: Not only the refresh produces: 1/ The loss of cell formatting 2/ the unability to do undo But also the conditionnal formating is lost, which was applied to the pivot table. To observe it. 1/ Go on the tab "Pivot table BUG and ANNOYANCES" 2/ select column B 3 Go to Format>Conditional Formatting > Manage (https://help.libreoffice.org/Calc/Conditional_Formatting) You'll noticed that the conditional formatting is still PRESENT, but no more applied.
In the light that this appear: 1/ independently of the file format (ods, xlsx or even if not saved, you can create a new document if the same state) 2/ there is no undo I suspect that the REFRESH actually calls a new constructor and NOT an UPDATE functions which must update the data content and leave the pivot table formatting as such, aka its css. Probably the update function is not coded at all. P.S. I reproduced the same conditions in MS-Excel 2000 (yes 2000), a pivot table refresh is indeed a data update and keeps the formatting.
confirming. btw: pivot table creates several cell styles starting with "Pivot Table ...". it will work, if you use these styles for formating.
Maybe the no undo can be a bug, but not that pivot table styles are applied con it's updated, this way it's how it works. As Oliver point same approximation can be done modifying pivot table styles. But seems the last solution it's broken e.g. modifying color background for a pivot table style works, but not number format that it's inherited from source cells. (LibreOffice 3.5.7.2 ) The last it's a regression from LibreOffice 3.3.0
If you say it is not a bug, then I say the function is can not be called "refresh". It's merely as calling a new constructor pivot table function with A. parameters being the formely selected source area B. dumping into the already target area sheet C. and overwriting any styling in the target. This is not the behaviour, as I mentionned in MS-Office, where a refresh is bound to a data update. >inherited from source cells. If you inherit from source cells, that you ought to inherit it's type, that's why for instance in my example I put one line as Header1 style, which is LO63rc2 does not do. Regarding inheritance of the source cells styling: 1/ ought to be an OPTION (for instance a thicker box), as it is not the default and expected behaviour when aggregating data. 2/ "refresh" behaviour ought to be consistent with other functions such as data filtering, sorting, etc. >Maybe the no undo can be a bug There is no undo, presumably because current coding flushes everything and so no undo is kept in the pile. Whether inheriting the source cells or keeping the existing styling, the currently implemented "refresh" functions seems either to lack in its calling arguments the styling and/or in its body it wipes out all styling.
*** Bug 133520 has been marked as a duplicate of this bug. ***
It's a very annoyng bug, see video on bug 133520 (duplicate). I want certain style for my tables. Maybe the boss tell you: I want that background here, that style here... And after changing one little ting everything is lost. The role of pivot table it's lost - it's a easy way to analyze and print data. But with loosing formating it's a long way in analyzing and printing data...
Marking as duplicate of bug 41645 *** This bug has been marked as a duplicate of bug 41645 ***
I just posted on bug 41645 That if my bug report is marked as duplicate of bug 41645 Notably, the bug 41645 does not mention the impossible undo.