[Libreoffice-bugs] [Bug 128329] WORK: 'incomplete autocalculations' - cooperation of autocalculate and shared formulas / formulae still broken in LO calc ver. 6.2.7.1 with *threaded* calculation on

[bugzilla-daemon at bugs.documentfoundation.org]

https://bugs.documentfoundation.org/show_bug.cgi?id=128329

b. <newbie-02 at gmx.de> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|1                           |0
             Status|NEEDINFO                    |UNCONFIRMED

--- Comment #9 from b. <newbie-02 at gmx.de> ---
hello @Buovjaga, 

steps: as the fail is sporadic and need not appear in multiple redos of similar
steps but only the one or other occasion i can't tell exact steps for
reproducing, just play around with data and check all results, 

file: you find one sample with a similar error i was able to catch in
tdf#125320, there with threaded of, i can't find more clue than 'wrong values
saved' in it, 125320 also has some 'steps' but redraws failed - to fail - ... 

i only catched the file with fail, not a version from before with detailed
steps until the fail :-( mostly such errors are spotted too late (by me), 

i spotted the errors during 'normal work', but in the meantime i've adjusted
that to avoid errors, actually with ver. 6.1.6.3, openCL and 'threaded' off,
thus it's unlikely i'll see one in the next time, 

(i cannot! work 4 weeks with the risk of data corruption in hope to probably
see a new error - which wouldn't give any more clue to the issue than
'autocalculate failed at this point' - )

for better forensics ... what about implementing an activity protocol and a
re-roll functionality, once somebody spottes an error he can save the file with
it's original values, and a 'script' about his actions, then that script can be
rolled through step by step, if the error shows up you have 'steps', if not you
have clue about 'weak hardware' or a sporadic fail in software ... 

as @erAck said: 'it always fails if sin(moonphase)+cos(mercury*round(sun))
touches a certain value' ... this is! rare but clearly too often! for a
reliable program ...

-- 
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/20201010/e3d7e5cb/attachment.htm>