Issue 54640 - Repagination of large documents takes minutes and/or crashes OOo
Summary: Repagination of large documents takes minutes and/or crashes OOo
Status: CONFIRMED
Alias: None
Product: Writer
Classification: Application
Component: formatting (show other issues)
Version: OOO 2.0 Beta2
Hardware: PC All
: P3 Trivial with 6 votes (vote)
Target Milestone: ---
Assignee: AOO issues mailing list
QA Contact:
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2005-09-15 10:55 UTC by postmorbid
Modified: 2013-08-07 14:38 UTC (History)
1 user (show)

See Also:
Issue Type: DEFECT
Latest Confirmation in: ---
Developer Difficulty: ---


Attachments
example for sxw that causes behaviour, most characters replaced by X (362.25 KB, application/vnd.sun.xml.writer)
2005-09-15 10:58 UTC, postmorbid
no flags Details

Note You need to log in before you can comment on or make changes to this issue.
Description postmorbid 2005-09-15 10:55:45 UTC
In a larger document (350+ pages, 1000+ footnotes), start editing the text
somewhere in the first half. If, for instance, you delete about a page of text
and the document ist 400 pages long, the status bar should simply show you 399
pages. Instead the page number increases to 420, 440, 460 and so on before
suddenly switching to 399 pages. While this is in process, performance of OOo
noticeably decreases (at least on slower systems (PIII 1.2 GHz, 512 MB RAM)) and
the text is flashing.

You can avoid this behavhiour by using "Update all", which works much faster
than the background paginiation (though it is also quite annoying to having to
do that).

Updating the pagination of large documents, however, crashes OOo (slower CPU) or
takes minutes (faster CPU). To reproduce:

- take a large document (300+ pages, 1000+ footnotes, as attached)
- decrease the font size for the main text
- repaginate the text with update all (works reasonably fast, a few seconds)
- now increase the font size to normal and do a repagination

This completely knocks out my PIII 1.2 GHz, 100 CPU load and nothing more
happening until a kill OOo (had it running for 6+ mins); on a faster CPU (2.8
GHz PIV) the last repagination takes about 5 mins 30 seconds (for the attached
document).

This makes editing / layouting such documents more or less impossible. This
seems to be a general problem which I encountered on several PCs with several
versions of OOo (1.5, 1.4, 2.0 beta etc.) with several different documents.
Comment 1 postmorbid 2005-09-15 10:58:13 UTC
Created attachment 29574 [details]
example for sxw that causes behaviour, most characters replaced by X
Comment 2 h.ilter 2005-09-19 15:33:58 UTC
Easier to reproduce with the upper part of the description. 
It has not crashed on my maschine but it's just an performance issue.
Comment 3 postmorbid 2005-09-19 20:18:50 UTC
> It has not crashed on my maschine but it's just an performance issue.

It may not have crashed your OOo or your machine (in fact it never crashed my
machine but it did crash OOo two times), but I am not sure if this is "just" a
performance issue. It probably is not much of a risk of data loss, but it makes
formatting such large documents more or less impossible.

If you have to wait for nearly 6 minutes on a - for normal office work - quite
powerful machine such as a PIV 2.8 GHz (or for 14 minutes on a PIII 1.2 GHz)
just to see how a decrease in font size affects your document, you can't use OOo
for formatting such texts. I have to admit that there are probably not too many
users who will need such large documents, but if you do, OOo is simply unusable.

Hopefully I do not come across as angry, but in my case this bug forces me to
switch to another programme, which is why I am not too happy to see this called
"just" a performance issue...
Comment 4 frank.meies 2005-09-20 10:39:19 UTC
FME: I can reproduce the performane problem, in fact it is quite annoying. I did
not find an obvious reason for the performance issue. Seems like the layout
algorithm needs some redesign. Considering the risc and effort and the fact that
this is not a regression, I have to shift the target to 'OOo later'.
Comment 5 jancs 2005-10-07 12:01:36 UTC
May be it is worth to make for the user an option to <Enable> or <Disable>
background repagination - i do not know about crashes, but it annoys awfully,
especially if the page layout is not important at all at the current stage of
work on document
Comment 6 postmorbid 2005-10-07 12:10:21 UTC
An option to disable background pagination would indeed be good, especially
since I guess it might be done much faster than fixing the bug itself.
Comment 7 bobkeyes 2005-12-08 10:35:43 UTC
I too have this same problem where it hangs if allowed to use background 
pagination.  Only I was having the problem with a master document upon loading 
with update all links.  It doesn't really do an update all links as the 
document loads and then continues with repagination much the same as a non-
master document.  This eventually hangs.  It wasn't until I saw this issue that 
I tried to load, respond no to update links and then when the file opens do an 
update all links and then it will completely load then entire document 
successfully.  Unfortunately I can't use master document because of another 
issue where master documents don't handle cross-references properly.  So I 
converted to using a single large odt file and inserted the sub-documents one 
at a time manually.  This works ok if an update all links is performed after 
each file insertion.  However, I made the mistake of trying to edit this large 
385 page document and caused a new page to be added which sent the document 
into background repagination from which it hung.

Also, I've noticed this background pagination, if it does work without hanging, 
does not produce the same output as an update all produces.  What happens with 
background repagination is that the page count is not stable and toggles 
between 381-385 like it can't figure out what it wants to do for sure.  This 
does not happen with an update all which produces a stable page count.  This 
pagination problem appears to be related to issue 58348 which seems to be 
referring to this same unstable page count.  But I'm not certain if these are 
the same problem or not or perhaps two parts of the same problem.  At any rate 
it makes it darn near impossible to create large documents with writer although 
master documents seem to work much better for this purpose if cross-references 
are not needed since the editing can be done in the smaller sub-documents and 
then just do one update all in the master document.
Comment 8 teeka421 2006-03-06 21:23:47 UTC
I have the same problem, but my document is only 2 pages. I created it with
OOo1.1 . It has tables, and a few graphics, but nothing extremely crazy. I
opened it, then tried deleting a few lines, slow - but accomplished it (CPU went
to 100% for a few moments) I then deleted a column, CPU went 100%, program
hanged. I'm wondering, if I make a new document using 2.0 - will that will fix
the problem (for now)?

Is anyone working on a workaround for this? Thx.
Comment 9 postmorbid 2006-03-08 06:56:10 UTC
It seems there are some more people having this kind of problem, some, like me,
with very large documents (the longer / the more footnotes, the worse it gets),
some with rather short ones. See, for instance,
http://www.oooforum.org/forum/viewtopic.phtml?t=31091&highlight=
Comment 10 bobkeyes 2006-03-15 20:21:06 UTC
I have a file that hangs every time I try to edit anythng in it.  In fact if I 
open the file and wait a minute it will hang.  If I open it and then 
immediately do an update all then it won't hang but as soon as I try to edit 
anything it will hang again.  As yet I haven't figured out how to recover the 
file.  I tried to copy it into another file but that has the same problem.  I 
tried to insert the file into a new document (a process that usually works to 
recover from this problem) but the new document also hangs.  The document is 53 
pages w/ lots of tables, footnotes, and cross-references.  I just downloaded 
and installed OO version 2.0.2 for Windows XP and the problem still persists.  
I would be willing to send the file but I don't want it posted because it 
contains some genealogy research I'm doing as well as personal information.  
But if the file might help in the debugging then I would send it directly to 
the appropriate investigator.
Comment 11 frank.meies 2006-03-16 06:25:06 UTC
FME->all: Most likely you are all encountering different problems. If you find a
crash/freeze/anything, you might consider filing a new issue. This way you can
assure that your special case is analyzed and dealed with.
Comment 12 mnasato 2006-11-14 23:51:32 UTC
My vote for an option to disable background repagination. Would be very useful
also for automation / scripting / headless operations.