[Libreoffice-bugs] [Bug 129199] AutoCalculate silent, deceptive fail. Indicative pattern of one edit step lag in AC update noticed.

[bugzilla-daemon at bugs.documentfoundation.org]

https://bugs.documentfoundation.org/show_bug.cgi?id=129199

--- Comment #10 from b. <newbie-02 at gmx.de> ---
Created attachment 156423
  --> https://bugs.documentfoundation.org/attachment.cgi?id=156423&action=edit
simpler_sheet_failing

hello R. Bingham, 

this bug ... remains me of a child play, one of a group is sent out, the others
define the rules, the one is called in again and has to ask questions to find
out the rules,  for each question he's given an answer app. for the previous!
question, 

thus LO is kidding with us? 

your second file has a shortcoming?, you are already confused about the
behaviour? or is it intentional that different solutions are correct? in row 96
you name results correct which are different from row 20 and 27. 

while not being able to solve the mystery logic- or technically, i found a
meta-level to explain: in the patch suspected by Oliver Brinzing erAck
announced: 'the results are somewhat... funny... '

https://gerrit.libreoffice.org/plugins/gitiles/core/+/eb678c01d35a485b9f79009c126e296d9e811d36/sc/source/core/data/formulacell.cxx

thus we get what he said ... ???

or, as Xerox said to David Kriesel: 'rtfm, it is written there, your fault' 

https://www.youtube.com/watch?v=7FeqF1-Z1g0, very good but in german. 

but ... may be ... your problem boils down to a simpler solution ... 

playing with attached sheet - which is a stripped down version of yours - (key
in values in A1), i see and estimate the following: 

- there is! circular dependency, if you c/p the line into a new sheet you get
err:522, oopsie: that applies only to specific versions? 

assumption: it's not 'seen' by calc in the original and my derivation because
we / you started with too complex formula construction, 

- the formula is 'solveable' in some steps, keying in the same value again,
likely it's needed more than two times, G1 is correct in every step, B1, C1 and
F1 narrow to the final result in every step, 

assumption: G1 only 'looks' correct, in real it's a sum of two deviating values
whose deviations cancel each other out, 

assumption: in your sheet some 'errors' are covered by the roundup and
rounddown functions, 

assumption: calc doesn't do any recursion by itself, that way N20 in your sheet
is lagging behind, 

- if you switch off 'iterations' you get err:522, 

assumption: calc doesn't do recursion 'in the sheet', only for display? or only
to check if err:522 is necessary? 

and now? ask the experts: 

- do we really have recursive / iterative calculation in the sheet? how to
activate? 

- if not ... what's the iterative setting in options good for? 

- and your sheet ... why didn't calc fire err:522? 

hope it helps ... reg. 



b.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/libreoffice-bugs/attachments/20191209/96990857/attachment.html>