Issue 123457

Summary:	Unacceptably slow in opening the particular 1800 pages ODT file (OOoFAQ)
Product:	Writer	Reporter:	Andrea Pescetti <pescetti>
Component:	open-import	Assignee:	Oliver-Rainer Wittmann <orw>
Status:	CLOSED FIXED	QA Contact:
Severity:	Normal
Priority:	P3	CC:	clarence.guo.bj, issues, jsc, oliver.brinzing, rainerbielefeld_ooo_qa, vitriol_vitriol
Version:	4.0.0	Keywords:	performance, regression
Target Milestone:	4.1.0
Hardware:	All
OS:	All
Issue Type:	DEFECT	Latest Confirmation in:	4.0.1
Developer Difficulty:	---

Description Andrea Pescetti 2013-10-10 06:54:15 UTC

A very large ODT file, which is an automatically generated collection of OpenOffice FAQs
(find a copy at http://people.apache.org/~pescetti/tmp/i123456/faq16.odt )
takes unacceptably long to open with version 4.x.

The performance problem depends a bit on the system, but opening with 4.0 is reported to take more than 20 minutes whereas on the same system 3.4.1 opens the document in 20% of the time.

A problematic cycle in 4.0x:
- Open the file (very slow)
- Save the file with another name (very slow)
- Reopen the new file (reasonably quick, expected speed)

[ Reported by Martello and Vitriol on mailing lists in Italian]

Comment 1 Rainer Bielefeld 2013-10-10 08:24:48 UTC

(a) Still Reproducible with  "AOO 4.1.0-Dev – English  UI / German locale - 
   [AOO410m1(Build:9750)  -  Rev. 1525917 - 2013-09-25]" on German WIN7 
   Home Premium (64bit)", own separate user profile, After 10 Minutes still 
   loading (progress bar 90%), after 15 minutes new progress bar, after 20 
   minutes still loading.
   Because of incomplete LCo selector (Bug 123063) no correct information 
   can be left.
(b) was ok with OOo 3.1.1, 30s for such a large document seems ok
(d) also Softmaker FreOffice only needs a minute or so.
(e) OOo 3.3.0 already is muc worse than 3.1.1, progress bar reaches 100% after 3 
    Minutes, new progress bar appears showing very slow progress, after 5 Minutes 
    document will be shown, so much bettern than 4.1.0-dev

Comment 2 Oliver Brinzing 2013-10-11 05:05:49 UTC

Comment 3 jsc 2014-01-09 15:36:33 UTC

I tried it with a local 4.1.0 dev version on my Mac and it took 3 minutes.

Comment 4 Oliver-Rainer Wittmann 2014-01-09 16:12:05 UTC

Jürgen and myself found out that more or less every paragraph contains a hyperlink without an URL. On saving the document these "empty hyperlinks" vanish and loading/saving time is as expected.

Taking over in order to try to tuned the ODF import of hyperlinks without an URL.

Comment 5 jsc 2014-01-09 16:18:10 UTC

some further info...
I saved the document under a new name which takes quite long (not measured but similar to loading). Closing the office and loading the new saved document took 30sec. See Olivers comment ...

Comment 6 Oliver-Rainer Wittmann 2014-01-14 11:34:49 UTC

solution in progress

improve ODF import:
- ignore <text:a> elements whose xlink:xhref attribute value (the hyperlink URL) is empty

improve ODF export:
- suppress export of hyperlinks which have no URL

Comment 7 SVN Robot 2014-01-14 13:10:52 UTC

"orw" committed SVN revision 1558026 into trunk:
123457: improve ODF filter - suppress import and export of hyperlinks without...

Comment 8 Oliver-Rainer Wittmann 2014-01-14 13:15:04 UTC

improved ODF filter in order to avoid import/export of hyperlinks without an URL

--> load and save of the given sample ODF document should take an acceptable amount of time

Comment 9 Clarence GUO 2014-04-03 06:00:27 UTC

Verified on trunk build rev. 1582712, fixed.
ENV: Win7 64 bit professional SP1