[This item is an out-of-band response to Clive that is included here with
the permission of Jeanne Sheldon. It provides an interesting case history,
especially with the Unicode wrinkle, and seems RISKS-worthy even if it
may seem like an old problem. PGN]
Here's a summary of what I've been able determine about the document.
The document was created in Word 97.
Word was set to allow "Fast Saves", which is a non-default setting that
performs incremental rather than complete saves. It is a feature
intended to speed the save operation. More information on fast save can
be found in Microsoft Knowledge Base articles:
Q71999 WD97: "How to Disable the FastSave Option in Word for Windows"
Q190733 WD97: "Opening Word Document in Text Editor Displays Deleted
Text"(this was first documented in Q113052 CREATED: 23-MAR-1994)
Q192480 WD97: "Frequently Asked Questions About 'Allow Fast Saves'"
The document was saved three times; the second save was to a different
filename. Because the second save initiates a second pass over the
document, Word was able to compress the Unicode so that it was readable
as ASCII characters and all incremental changes that were Fast Saved
were collapsed. The first letter was then deleted and the letter to Dr.
Page was composed. A single save was then performed to a local
(non-network) drive using the same filename. Because "Fast Save" was
enabled, the deleted text stream was identified but not actually
deleted. Because a single save is a single pass and Unicode compression
requires a second pass, the text remained as uncompressed Unicode. On
Unicode compression, see: Q168967 "File Size Twice as Big When Compared
to Earlier Version." While a non-Unicode aware tool would be unable to
read the second set of text (the letter to you), it is actually quite
readable on a Unicode-enabled text reader.
Extra notes: The document contains a unique identifier, indicating that
the version it was authored on did not include the fix which removes
that identifier. See Q222180 Unique Identifiers and Microsoft Office
97 Documents.
The document title, under properties, is generated automatically from
the first line of the document on the first save. It is not
subsequently updated, so it may contain text that is no longer in the
document.
Comprehensive information on the topic:
Q223790 How to Minimize Metadata in Word Documents.
From Word 97online documentation:
The difference between a fast save and a full save
If you select the Allow fast saves check box on the Save tab in
the Options dialog box (Tools menu), Word saves only the changes to a
document. This takes less time than a full save, in which Word saves the
complete, revised document. Select the Allow fast saves check box when
you are working on a very large document. However, a full save requires
less disk space than a fast save. If you are working on a document over
a network, clear the Allow fast saves check box. Fast saves cannot be
performed over a network.
You should do a full save in the following situations:
* Before you share a document with other people
* When you finish working on a document and save it for the last time
* Before you begin a task that uses a lot of memory, such as
searching for text or compiling an index
* Before you transfer the document text to another program
* Before you convert the document to a different file format
Note: If you select the Always create backup copy check box on the Save
tab in the Options dialog box (Tools menu), Word clears the Allow fast
saves check box, because backup copies can be created only with full saves
... Clive, thank you very much for the time and effort that you have put
into this. Although the Word setting that caused the document to be
created in such a manner goes back to a time when electronic document
exchange was not the norm (and, over the past 7 years, much effort has
gone in to attempting to assure that private information is not
accidentally included) it is humbling and daunting to realize once again
how difficult it is to correct the mistakes of past versions with software
patches, bulletins and product documentation.
Jeanne Sheldon, Microsoft Corporation ["Jeanne Sheldon" via risks-digest Volume 21, Issue 43]
0:00
#
G!
