Issue 119232 - Took too long time to open 400K .xlsx files.
Summary: Took too long time to open 400K .xlsx files.
Status: ACCEPTED
Alias: None
Product: Calc
Classification: Application
Component: open-import (show other issues)
Version: 3.4.0
Hardware: All All
: P2 Normal (vote)
Target Milestone: ---
Assignee: AOO issues mailing list
QA Contact:
URL:
Keywords:
Depends on:
Blocks: 120823
  Show dependency tree
 
Reported: 2012-04-16 08:42 UTC by Li Lin Yi
Modified: 2023-08-04 14:23 UTC (History)
8 users (show)

See Also:
Issue Type: PATCH
Latest Confirmation in: 4.1.7
Developer Difficulty: ---

Attachments
Sample file. (412.77 KB, application/vnd.openxmlformats-officedocument.spreadsheetml.sheet)
2012-04-16 08:42 UTC, Li Lin Yi 		no flags Details
remove the oox::xls::CellBlock cache (20.14 KB, patch)
2012-10-17 14:27 UTC, hdu@apache.org 		hdu: review? (leiw)
 Details | Diff
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this issue.
Description Li Lin Yi 2012-04-16 08:42:10 UTC 
Created attachment 77450 [details]
Sample file.

Took 1:14 minutes to open the attached sample file.

It took only 11 seconds to open the same file on OO3.3.
Comment 1 Oliver Brinzing 2012-04-16 10:41:04 UTC 
confirming on vista 32bit with aoo3.4 r1325589
Comment 2 hdu@apache.org 2012-10-10 14:05:53 UTC 
This regression was to be introduced by the CWS dr78:
  #164376# use XCellRangeData to increase import performance
as it indirectly causes a lot of reformatting operations.

Reverting the relevant changes or even just disabling the mechanism with a one-liner gets the performance of loading the document back to OOo33 levels. It would be interesting to have access to #164376# to identify use cases that benefit from the CWS changes, but this closed bugtracking system has apparently been lost for good.

If the new AOO automated PVT tests show that just removing/disabling the changes helps without too much impact, then this should be done.
Comment 3 hdu@apache.org 2012-10-17 14:27:30 UTC 
Created attachment 79789 [details]
remove the oox::xls::CellBlock cache

The CellBlock cache introduced in CWS dr78 caused a major performance regression because it resulted in the document being reformatted over and over during XLSX import. As there is no simple fix to avoid the reformattings caused by the CellBlock cache it should be removed altogether.
Comment 4 hdu@apache.org 2012-10-17 14:32:07 UTC 
Comment on attachment 79789 [details]
remove the oox::xls::CellBlock cache

The patch reverts parts of the commit from CWS dr78 on 2011-02-25
"dr78: #164376# use XCellRangeData to increase import performance"
but keeps the cleanups.
Comment 5 Kay 2016-07-17 23:09:34 UTC 
This change was not included in 4.0.0. However, it may be still needed. We need some testing on this.
Comment 6 Marcus 2017-05-20 11:01:01 UTC 
Reset assigne to the default "issues@openoffice.apache.org".
Comment 7 oooforum (fr) 2020-01-04 13:45:54 UTC 
Opening takes ~30 sec. with AOO 4.1.7 under Win10.
Maybe decreased with better computer performance (hardware or system). 

According to comment 4: set as PATCH
Comment 8 damjan 2023-08-04 02:05:08 UTC 
I had high hopes for this patch, but it does not improve performance on the attached file (which already loads in only 14 seconds) or a number of other performance bugs I tested:
Bug 124022
Bug 123429
Bug 120619
Bug 123919