Created attachment 92050 [details] A spreadsheet with the sequential series of automatic sums going berserk every 3/4 instances Problem description: when you create a formula using the automatic sum tool, or the function wizard, cell ranges created by default or by drag and drop are often wrong and result in error messages. When using the automatic sum, if data are nearby their cell ranges are usually inserted by default. Now however this insertion if very often - but not always - totally wrong. Steps to reproduce: I'm still struggling to find out a fully deterministic, reproducible mechanism for this, it seems to happen about 1/3 of the times. The file I am attaching shows a simple mechanism of sum insertions and the ugly outcome: - type 1 in A1 and move to A2 - insert an automatic sum in A2, then press return and move to A3 - insert an automatic sum in A3, then press return and move to A4 - repeat until you reach A30 The result should be a series of sums of the above column, each time with one more value being added; instead it is a total nightmare as in the attached file, with many autofill giving things like SUM(A7:#RIF!A7) (see attached image). When using some drag and drop cell range definition, like in the function dialog, the same thing happens, sometimes in place of a thing like A7:#RIF!A7 you get a A3:Foglio3.B15 instead of an A3:B15 (which is wrong as well). And when you select a single cell range this is wrongly indicated repeating twice the single cell, so that A1 becomes A1:A1. Current behavior: All this is totally critical to use Calc with functions and formulas, making 4.2 RC2 substantially unusable. Having realized this was happening I immediately uninstalled the 4.2 RC2, the risks of serious errors are too high. Expected behavior: Operating System: Windows 7 Version: 4.2.0.2 rc Last worked in: 4.1.4.2 release
I forgot to write explicitly that all this erratic behaviour never happened in 4.1.4. Also, in the attachment there are some images showing partial screenshots of inappropriate cell range fills.
I've seen this in rc1, not (yet) in rc2, Andy. Do you have a sample file + steps pls? thanks a lot! Cor
Ehm, I am not sure to understand what are your requests.... I have already attached a file where the wrong autofill cell range are obtained with a precise and simple procedure in column A2, described in the original bug report: - type 1 in A1 and move to A2 - insert an automatic sum in A2, then press return and move to A3 - insert an automatic sum in A3, then press return and move to A4 - repeat until you reach A30 As you can see in the file, starting from A4 the automatic sum autofill range goes wrong and this happens again a few times more before line 30 If this is not what you mean, I am sorry but I must ask to better specify what you would like me to upload. All problems were observed in 4.2 RC2. I actually did not installed RC1 due to lack of time. Thanks for your help, anyway
I am sorry to bother, and of course I am in no position to complain, but IMHO not addressing this bug asap (I have just checked the fixed bug list for RC3) means that any intensive and systematic use of calc is not possible without high risks for the reliability of any computation performed. Could anyone please confirm that this problem will be in RC3 just as it was there in RC2? If this is the case, I'd rather actually wait for a subsequent release before any further testing.
Created attachment 96543 [details] OS Windows, LibO-4.2.3rc2_install This specific error. It occurs only in Windows. I use the Windows XP (32 and 64 bit). I confirm this bug for programs: - LibreOffice 4.2.3.2 (ID: 7c5c769e412afd32da4d946d2cb0c8b0674e95e0); - LibreOfficeDev 4.1.6.0.0+ (ID: b6d9aa63ab63f3236712b2b68155e6b0d129a68); - LibreOfficeDev 4.3.0.0.alpha0+ (Build ID: 325ecb1d0511c6c02d863eb760c2d0521f0b6c19, TinderBox: Win-x86@39, Branch: master, Time: 2014-03-25_02:26:28) -- P.S. Lubuntu-13.10 (amd_64), LibreOffice-4.2.2 - OK
Created attachment 96544 [details] OS Windows; LibO-4.0.6_portable This is a regression to LibreOffice 4.0.6.2 (ID: 2e2573268451a50806fcd60ae2d9fe01dd0ce24)
(In reply to comment #5) > Created attachment 96543 [details] > OS Windows, LibO-4.2.3rc2_install > > This specific error. It occurs only in Windows. I use the Windows XP (32 and > 64 bit). I confirm this bug for programs: > - LibreOffice 4.2.3.2 (ID: 7c5c769e412afd32da4d946d2cb0c8b0674e95e0); > - LibreOfficeDev 4.1.6.0.0+ (ID: b6d9aa63ab63f3236712b2b68155e6b0d129a68); > - LibreOfficeDev 4.3.0.0.alpha0+ (Build ID: > 325ecb1d0511c6c02d863eb760c2d0521f0b6c19, TinderBox: Win-x86@39, Branch: > master, Time: 2014-03-25_02:26:28) > -- > P.S. Lubuntu-13.10 (amd_64), LibreOffice-4.2.2 - OK -- Sorry, I was wrong: LibreOfficeDev 4.1.6.0.0+ (ID: b6d9aa63ab63f3236712b2b68155e6b0d129a68) works correctly.
There is a bit of confusion, so to clarify, what Andy refers to as "automatic sum" is the icon that shows the Greek sigma sign located up on the toolbar between the Function Wizard icon (fx) and the Function icon (=).
This works fine for me on Windows as well, using the latest master. My outdated 4.2 build still shows this bug. Let me update that branch and see if the latest 4.2 fixes this as well.
FYI, Try turning off OpenCL support in the formula calculation settings. That should give you a workaround.
Just verified that this is no longer an issue with the latest 4.2 build. I tested it with Version: 4.2.5.0.0+ Build ID: c4e10ad83fe1fa92a115327c1909218f5dc8c8bd Not sure when this got fixed. 4.2.4 may or may not have this issue, but I didn't check with that version.
(In reply to comment #11) > Just verified that this is no longer an issue with the latest 4.2 build. I > tested it with > > Version: 4.2.5.0.0+ > Build ID: c4e10ad83fe1fa92a115327c1909218f5dc8c8bd > > Not sure when this got fixed. 4.2.4 may or may not have this issue, but I > didn't check with that version. @Kohei: You have made the patch for another bug, which is similar to this. Unfortunately, I do not remember the bug’s number. The patch has been included in the LibO_Master and LibO_4.2.5.0.0+ ~ two weeks ago. It’s possible that there is no bug now for this reason.
Migrating Whiteboard tags to Keywords: (PossibleRegression) [NinjaEdit]