Bug 115749

Summary:	Deleting a (very) large number of sheets is slower than it used to be
Product:	LibreOffice	Reporter:	Telesto <telesto>
Component:	Calc	Assignee:	Not Assigned <libreoffice-bugs>
Status:	RESOLVED WORKSFORME
Severity:	normal	CC:	ilmari.lauhakangas
Priority:	medium	Keywords:	bibisectRequest, perf, regression
Version:	6.1.0.0.alpha0+
Hardware:	All
OS:	All
Whiteboard:
Crash report or crash signature:		Regression By:

Description Telesto 2018-02-15 15:23:00 UTC

Description:
Deleting a large (very) number of sheets is slower as it used to be

Steps to Reproduce:
1. Open attachment 134550 [details]
2. Right click the second sheet & select All
3. Deselect the first sheet (ctrl+click)


Actual Results:  
Deleting takes around 8 minutes with LibO6.1

Expected Results:
Something around 3,5 minutes, similar to LibO4.4.7.2 (or better)


Reproducible: Always


User Profile Reset: No



Additional Info:
Version: 6.1.0.0.alpha0+
Build ID: 3c913c3844acae8ee0d80ab174133bdc7677efea
CPU threads: 4; OS: Windows 6.3; UI render: default; 
TinderBox: Win-x86@42, Branch:master, Time: 2018-02-14_00:19:27
Locale: nl-NL (nl_NL); Calc: CL

but not in
Versie: 4.4.7.2 
Build ID: f3153a8b245191196a4b6b9abd1d0da16eead600
Locale: nl_NL


User-Agent: Mozilla/5.0 (Windows NT 6.3; WOW64; rv:52.0) Gecko/20100101 Firefox/52.0

Comment 1 Xavier Van Wijmeersch 2018-02-15 20:58:27 UTC

confirm with master build today 7min

but more than 500 sheets, its not a race to win we wont, i think

Comment 2 Telesto 2018-02-15 21:12:47 UTC

(In reply to Xavier Van Wijmeersch from comment #1)
> confirm with master build today 7min
> 
> but more than 500 sheets, its not a race to win we wont, i think

It's about the comparison. LibreOffice 5.4 deletes the 588 sheets in around 200 seconds. LibO6.1 needs around 480 seconds. 

This is an excessive example. I'm aware of that. However the performance decrease is obvious and significant. Which should happen in smaller cases too.

Comment 3 Buovjaga 2018-06-02 14:44:18 UTC

Results from Win 10:
6.2: 3min 19s
6.0.4: 3min 27s
5.4.2: 3min 20s
5.0.2: 2min 59s (could not jump to first sheet, control was missing)
4.4.7: 2min 48s

I don't know, who would want to spend ages bibisecting this especially with such gradually progressing results. I should probably do a callgrind.

Version: 6.2.0.0.alpha0+ (x64)
Build ID: a8f8cf72b2b9e912dc4a5aebef55d9b2c0969462
CPU threads: 4; OS: Windows 10.0; UI render: default; 
TinderBox: Win-x86_64@42, Branch:master, Time: 2018-05-30_15:31:15
Locale: fi-FI (fi_FI); Calc: group threaded

Comment 4 Telesto 2018-06-02 20:38:37 UTC

(In reply to Buovjaga from comment #3)
> Results from Win 10:
> 6.2: 3min 19s
> 6.0.4: 3min 27s
> 5.4.2: 3min 20s
> 5.0.2: 2min 59s (could not jump to first sheet, control was missing)
> 4.4.7: 2min 48s
> 
> I don't know, who would want to spend ages bibisecting this especially with
> such gradually progressing results. 

Sounds more like a WFM: Report was about Version: 6.1.0.0.alpha0+ being slow (8 minutes). It's back to 'normal' with 3 min 30 seconds.