[Libreoffice-bugs] [Bug 114482] Circular reference accumulator, err:523 or miscalculation

[bugzilla-daemon at bugs.documentfoundation.org]

https://bugs.documentfoundation.org/show_bug.cgi?id=114482

--- Comment #13 from b. <newbie-02 at gmx.de> ---
still bug: iteration is performed twice on initial set of 

A1: '=A1+1' or similar, (iterations: 1 minimum change >=1), result: '2', 

still bug: with above formula iteration doesn't star with former value of the
cell, but with zero, 

still bug: on subsequent recalcs (strg-shift-F9) iteration is performed once if
set to 1, twice if set to 2 or above, but not more than twice even when set to
1000, 

the original question works nearly as expected - as requested by the OP - with: 

Version: 6.2.8.2 (x64)
Build ID: f82ddfca21ebc1e222a662a32b25c0c9d20169ee
CPU threads: 8; OS: Windows 6.1; UI render: GL; VCL: win; 
Locale: de-DE (de_DE); UI-Language: en-US
Calc: threaded

as well with the RAND() formula as with fixed values (which is easier to follow
/ easier to see how many steps are performed, use ctrl-shift-F9 or new input to
the cell to trigger new calculation)

fixed? intentionally? by chance? 

one thing left: if you start with '0' in A1, and enter the formula '=A1+1' in
A1 with iterations set to 1 and minimum change set >= 1 i'd expect a result of
'1'
... funnily it's '2', on subsequent calculations triggered by hard recalc calc
steps through the numbers one by one, just the start is deviating, 

it doesn't matter with which value you start, if you replace any numerical
value in a cell with a formula like '=A1+x' or '=SUM(A1)+x' or '=A1+B1' or
'=SUM(A1:B1)' (x value to choose) the first iteration will not produce the
expected result but the double value of the summand to add (in the samples x or
B1), 

it looks as if the iteration doesn't start with the former value of the cell,
but will instead use the value of x or e.g. B1 as a start value and perform the
iteration once, or use 0 as start value and perform the iteration twice, 

similar applies for '-' calculations, while multiplications and divisions start
with a zero value? 

new bug? 

one other thing irritating: the setting of 'Minimum change:' in the options
looks more like working as the *maximum change* above which calc switches to
Err:523, 

enhancement request? 

good to know: when the formula / iteration results in Err:523 the calculation
is performed anyway, and the resulting value is stored and is used for the next
iteration, Err:523 is inherited by dependent cells, and cleared once the
conditions for the first cell are ok again, 

and even one more 'uncertainity' ... 

for A1: '=A1+1' the version below starts with two times adding 1 to zero, and
on subsequent recalcs adds 1 if iteration is set to 1, and 2 if iteration is
set to any higher value ... ??? there is still work left in this area ... 

Version: 6.5.0.0.alpha0+ (x64)
Build ID: 9ab43aebad67383057d2cc3f754ce2193fa78b4e
CPU threads: 8; OS: Windows 6.1 Service Pack 1 Build 7601; UI render: default;
VCL: win; 
Locale: de-DE (de_DE); UI-Language: en-US
Calc: 

applies to other versions too

-- 
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/20191217/e7cf9f88/attachment.htm>