Master Document

[thboyd]

This is a continuation of a post that was originally in a thread named SLOW!!, but since I'm changing the topic somewhat I've started a new thread.

In an attempt to get better performance with a 230 page text document that contains about 200 illustrations, I converted the document to a Master Document. I've been using this new master document for about two days now and find that it is not satisfactory.

Document description: 240 pages, 12 subdocuments, three tables of contents (in the master document), one index (also in the master document). The subdocuments range from about 4 pages with one illustration to about 30 pages with about 30 illustrations. The master document as well as all of the subdocuments are linked to the same template and all formatting is defined by styles (page styles, text styles or character styles).

First, I should point out that this is all done with graphics and objects viewing disabled. Here's what I've found:

1. It is very slow to update. If I make even a trivial change in a subdocument (like changing a cross reference) it takes from 5 to 10 minutes for the master document to update (This is determined by monitoring cpu utilization in Activity Monitor, since NeoOffice gives no indication that it's doing anything). This seems to include repaginating the entire document because during this period the Activity Monitor reports any where from 2% cpu utilization to 100% utilization and the page count on the master document changes, often showing significantly more pages that actually exist in the document. I think this is because there are sometimes many blank pages inserted before illustration. In the subdocuments the pictures are linked rather than embedded.

2. Some cross references are wrong. The text often contains cross references to illustrations and the illustration numbers are occasionally wrong, e.g. Figure 25 instead of Figure 20. Sometimes the hot link is linked to the correct illustration even though the number is wrong, but not always.

3. Some styles are not displayed correctly. Specifically, I am using 5 Heading styles. Heading2 and Heading3 are sometimes (not always) displayed with an indent even though inspecting the style shows that there is no indent defined. Quitting and restarting NeoOffice sometimes corrects this and sometimes reattaching the template will correct this, but neither of these always works.

4. Updating all does not always update all. Sometimes I have to specifically update fields in order for for the illustration number fields to be numbered correctly.

My question is: Do other people have similar problems with master documents?

Finally, I'm using NeoOffice 3.0.2 Patch 1 on an iMac with a 2.4 GHz Intel Core 2 Duo, 4 Gbytes of RAM and OS 10.6.2.

[pluby]

When I read your post, I think the key thing that you are seeing is that our suggestion in the other thread to use a master document does not yield any appreciable speed improvement over a single massive document.

It seems that there are lots of other issues with using a master document so I think we need to go back to the speed issue. The signature in your post indicates that you have a fast machine so I do not think the hardware is the issue.

You have also turned off image display so at this point so we are running out of the obvious things to try so I really would like to see the original single massive document and see if I can reproduce what you see as in both NeoOffice and OpenOffice.org. Do you have the document on a site that we can download it from?

Patrick

[ovvldc]

Another common cause of slowdowns is reflow. If there is a long section of text without page breaks, NeoOffice has to do a lot of size calculations to make sure the pararaphs are placed where they should be. If you have page breaks regularly, these reset the reflow for part of the text, so the reflows might be faster.

But I cannot judge if that would help unless I see the document..

Best wishes,
Oscar

[rays]

I was taking a look at this in NeoOffice Help accessed from NeoOffice>Preferences>NeoOffice>Memory and clicking on Help:

<code>Graphics cache
The graphics cache saves the graphics contained in a document in your computer's main memory. This means that the attributes of a graphic stored in the cache do not have to be re-calculated if you return to the page containing the graphic after scrolling through a document.

Use for NeoOffice (MB)
Specifies the total cache size for all graphics.

Memory per object (MB)
Specifies that objects which are larger than the selected megabytes will not be placed in the cache.

Remove from memory after (hh:mm)
Specifies the time that each graphic remains in the cache in hours and minutes.</code>

This got me thinking. If you can establish what size an average image is in your document, you could use that as a basis for calculating a better optimization of NeoOffice for your particular situation.

For example, setting Memory per object lower than your average image size will prevent NeoOffice from being able to cache it, meaning such images would have to be constantly re-calculated. This would be a bad thing. So push this up well above average image size.

Since you have 4GB of memory, you could greatly increase the "Use for NeoOffice" setting, maybe at 10 or 20 times your average image size or even more. There's no point in skimping in this area when the memory is otherwise idle...

Just a thought.

[thboyd]

The settings that I've been using are:

Graphics cache:
Use for NeoOffice: 256 MB
Memory per object: 16 MB
Remove from memory after: 00:20
Cache for inserted objects:
Number of Objects: 20

256 MB seems to be the maximum allowed for the Use for NeoOffice setting. If I increase it, it resets to 256.

I can increase the Memory per object, but I don't think that I have any images that exceed 15MB

I assume that the cache for inserted objects is not material, since all of my images are linked, not inserted. Is that right?

[rays]

That all seems fine. I wasn't aware of the apparent 256MB upper limit.

Are these images photographs? If so, what type of printing process will be used for the final publication? In-house colour or greyscale printing on an office printer? Or something requiring a higher image resolution? If so, what resolution?

The reason I ask is that there is significant optimization to be gained by re-sizing all images to the absolute final dimensions they will have on the page at the correct resolution appropriate to the printing process.

If you have not already gone through that step with your images, you may be surprised at how much they might be downsized without any loss of quality at the end of the process.

In other words, you should not have any images in your final document which you have cropped or re-sized using the tools provided in the office package. These only affect the displayed image but the underlying image remains the same - often far too great - file size, requiring massive amounts of calculations and processing every time they are rendered, I believe.