calc bug regarding comments? - pls. help with testing

hello @all, in particular to people interested in the performance of calc,

please perform this simple test:

start with an idle computer and an empty worksheet,

don’t do any testing while productive work is unsaved, your system may freeze or crash!!!

put ‘test’ in A1, and add the comment ‘test1’,
put ‘test1’ in B1, and add the comment ‘test’,
copy A1 to B2, and B1 to A2

this creates a 2x2 ‘checkerboard pattern’ of data and comments, you can click around in it, no problem, open task manager to check system load,

now expand this pattern to rows 1:100, ctrl-c - ctrl-v, and click around in that - my system shows slightly increased system load,

now expand this pattern to a 100x100 field (A1:CV100), ctrl-c - ctrl-v, and click around in that - my system shows massive!!! increased system load (about 12% of a powerful xeon, enough to push the fans from off to level two), and switches to ‘not responding’ for some time on load, save, copy and other actions. :frowning:

estimate - or try :wink: - what will happen if you expand to 1000x1000 … :frowning: :frowning: :frowning:

cross-check without the comments, on my system ‘nopro’ even with much biger amounts of data,

boundary conditions: some of the problem may depend on ‘saving auto recovery info’ being turned on, and / or will come up / show increased impact after the first autosave has taken place, also the problem may depend on other circumstances (e.g. i use the touchpad of the laptop, may be it’s different with mice), that’s why i’m asking here,

this test is ‘against’ a very old (5 years) unresolved bug:

https://bugs.documentfoundation.org/show_bug.cgi?id=76324

in the discussion there it’s somewhat unclear if the bug is a bug, and if it’s still in, or solved, or covered ???

imho:

  • the bug is a massive! and ‘trapping’ bug,
  • it is! still present,
  • it should be killed,
  • the community at ‘bugs’ needs some ‘field testing’ to identify the source of the evil,
  • the community at ‘bugs’ could use some motivation to take this bug seriously,
  • users should be warned before potentially valuable work comes to an unworkable state,
  • users should be warned before plenty of them gather frustration, waste time in identifying and testing, and so on,

P.S. this problem showed up during normal work with versions 5 and 6, and led back to a problem reported for a 4 version, it may be and will be normal that it occurs different in different versions, it may be helpful to distinct / compare between versions, thus pls. give info about that, my system: lenovo thinkpad, xeon, 16gb, ssd disk, win7pro (x64), LO 6.3.0.0alpha1+

P.S.II normally problems / questions like this have a better place in ‘bugs’, - i know that! - i’d write here additionally to get some help in analysis and some help to end the ‘delaying’ there,

@Hrbrgr: funny comments :slight_smile: ‘A: see B’, ‘B: see A’, programmatically ‘endless loop’? if the handling of comments in calc is done in a similar way it’s inescapeable to produce performance issues …

fun aside: yes, these two questions are ‘related’, but! they are targetting two very different aspects of the problem,

  • one is to gather info about shortcomings in common and help to avoid them,

  • the other is to get this bug tested and killed out,

reg.

b.

as of now - 2019-11-17 - this behaviour is massiely reduced in ver. 6.2.7.1 win(x64), while it impressively slows down the parallel linux version.

subsequent windows versions show enhanced impact again,

i’d like to get some people to test the behaviour on their systems and give feedback, eiter here or at bugs.libreoffice.org.

thanks a lot