![]() |
"The most comprehensive and current FrameMaker resource, outside of the Adobe site" (Keith Soltys, Essential Resources for FrameMaker Users) |
|
|
|
|
|
|
|
|
|
![]() |
|
![]() |
|
![]() |
|
![]() |
|
From Redaction of Confidential Information in Electronic Documents, an Adobe technical note (PDF: 614K, 13 pages, March 2006): "Both Word and PDF documents can carry metadata information about the document, such as author, subject, keywords, and title. The author may be unaware of metadata generated by the application, and it may not be apparent unless the user knows where to look for it. ... The checkbox Convert Document Information controls the conversion of Microsoft Word metadata to PDF and is checked by default." This is followed by a recommendation to turn off this feature.
The importance of the Title field, and optionally other document information fields, was discussed in several Hmmm... items, together with numerous examples of how these fields may contain useless/ridiculous values ("This is a placeholder. Final title will be filled later") in PDFs. If the Convert Document Information option is turned off, as recommended by Adobe, then the Title (and other fields) will have to be entered manually every time the file is converted to PDF. It is true that sensitive/internal metadata should not be present in a PDF, but why recommend disabling a good feature instead of educating users as to how this is best handled as part of the authoring process in Microsoft Word? Adobe, as the parent of this useful feature, could be expected to work towards establishing standards for useful yet non-sensitive metadata, instead of throwing the baby out with the bath water.
Taking the logic behind the recommendation made in the technical note further:
Furthermore, the technical note does not address a real issue:
PDFs are commonly created from Microsoft Word by printing to
the "Adobe PDF"/Distiller printer drivers directly
(without PDFMaker), in which case the Title field shows the source
file name/path and the Author field shows the log-in name --
which are potentially far more sensitive than Document Information
fields set specifically by the document author (see examples). The Title is handled the same
way, even when the "Add Document Information" field
is turned off in the printer driver properties.
When using text shading (Microsoft Word) or artificially "bolded" fonts (Adobe FrameMaker), resulting PDFs have multiple instances of the same text, disrupting text-related functions such as search, copy/paste or "read out loud" (with some variations depending on the authoring tools, drivers, and Acrobat/Reader versions being used).

What is an e-book? "Sometimes presented as Ebook
or EBook, the term refers to a book that is available in electronic
format. Usually eBooks are available in Adobe PDF or eBook Reader
format, or in Microsoft's LIT format, but there are many other
formats available. A good eBook uses the technology effectively,
with tables of contents that link to the correct chapters and
search capabilities. Links to Web sites and other files also
can be included." (www.bookzonepro.com, Glossary)
The e-book (PDF) included with
Microsoft Manual of Style for Technical Publications, 3rd edition
could have been a great opportunity to demonstrate the added
value of an interactive e-book, and the subject matter and technical
character of the book certainly lends itself to interactivity.
What a disappointment to discover that it is yet another sub-standard
PDF promoted as an e-book. For starters, the table of contents
has no links. Likewise, the index entries are not linked. Nor
are the numerous cross-references throughout the book. In fact,
this PDF has a total of ... one link (on page 318), which happens
to be a bad link.
The PDF was authored with FrameMaker 7.0. As there are traces of FrameMaker-derived interactivity (e.g. destinations, bookmarks, article threads), it is highly probable that the links were removed after the PDF was created.
[Note: the PDF replaces a compiled HTML help (CHM) provided with the 2nd edition, which was fully-interactive and easy to use. As the PDF was created as a single PDF for the entire book, it does not utilize the potential benefits of Acrobat's search mechanism. Had this been a collection of separate PDFs, each with its own unique title (and possibly keywords), Acrobat Search could have yielded results similar to or better than those of the CHM search. With the "entire book = single PDF" approach, the reader is confined to clueless, serial jumps in a long document.]
The Open Service Event Manager User Guide (PDF: 1.9MB) has 50 bad links out of 908 links.
Of the remaining 858 links, there are approximately 300 unintentional
links. In addition, many of the cross-references are a combination
of two links, one pointing to the heading number and the other
to the heading text. When you consider these factors, the ratio
of bad to good links is more likely in the range of one bad link
out of every five links (most of the links in the Glossary are
bad, for example).
[In FrameMaker-authored PDFs, a common reason for random bad
links is having "Create Named Destinations for All Paragraphs"
turned off in FrameMaker's PDF setup, Links tab.]
Unintentional links are often created in PDFs when cross-references
are used in the source files for the purpose of retrieving text,
for example a book title or release date. When such cross-references
are placed in the header/footer and repeated on all pages, their
presence becomes significant, and adds a burden to the file size
without any benefit.
![]()
FrameMaker-specific: When using FrameMaker to author PDFs, it is highly recommended that you use variables and not cross-references for text retrieval. There are a number of reasons for this:
The Adobe FrameMaker 7.2 "Structure Application Developer's Guide" PDF has several diagrams which simulate FrameMaker's Structure View window, where elements are shown as bubbles (the one below is on page 168). The quill-like effects (display and print) are not related to drawing by hand, but to a new setting in Distiller 7. When "Convert smooth lines to curves" (Advanced tab) is turned on (and it is turned on by default in the "Standard" and "Smallest file size" options), FrameMaker rounded rectangles are warped (this has to do with the way FrameMaker outputs its graphic objects to PostScript, splitting basic shapes into multiple segments).

When the View >
Read Out Loud function in Acrobat/Reader 7 is activated in the
Adobe FrameMaker 7.2 "User Guide Supplement" (52-page
PDF provided as part of the product), Acrobat's computer voice
says "Warning! Empty Page!" for about half of the pages
(1, 2, 12, 22, 24, 28, 32-45, 47, 48, 52).
In the Adobe FrameMaker 7.2 "Structure Application Developer's
Guide" (619-page PDF), the "Empty Page" warning
is issued for all pages.
This PDF has a total of 109 bad links (in pages 30, 134-139,
202, 230, 593-600). Page 15 has 123 valid links even though it
is really empty (link areas are displayed when Acrobat's
Link tool is selected). The bad links are likely the result of
having "Generate Named Destinations for All Paragraphs"
turned off (FrameMaker); the superfluous links are likely the
result of page replacement (Acrobat).

"Making visually rich, personalized content reliably available anytime, anywhere, on any device" is the essence of Network Publishing. Yet Adobe's presentation of this concept, The Next Wave: Network Publishing (PDF: 801K), is a bitmap-based PDF. In such PDFs, text is not accessible (no search or copy), display/print quality is poor and typically the file size is larger than it should be.
Adobe, everywhere
you look: The first and last page (identical) of the
presentation consist of the Adobe logo and the "everywhere
you look" phrase as text -- zoom in and the
quality of text is not affected, and you can also locate the
text by searching.
"Adobe, everywhere you look" is also present in all other pages, in a bitmap form, at the bottom-right corner. Because of the poor visual result, an opaque white rectangle with the Adobe logo (in a high-quality vector format) was placed on top of the "Adobe, everywhere you look" bitmap. When paging down, the "hidden" items show up briefly on screen (even though not visible in printed output). To see what's below the vector logo and the white rectangle, use Acrobat's TouchUp Object tool.
PS. When activating the View > Read Out Loud function (Acrobat/Reader 6 or higher) for the entire PDF, Acrobat has the following to say (in its synthesized voice): "You look trademark everywhere. BBC. BBC. BBC. BBC. BBC. BBC. BBC. BBC. BBC. BBC. BBC. BBC. BBC. BBC. You look trademark everywhere."
When PDFs are searched on the web, the PDF title (if present) is displayed in the search results. In the absence of a title value, content from the first paragraph is displayed (sample search, pointing to a sample PDF file).

Generic titles assigned to PDFs by default are counter-productive, and far worse than not having any title -- as shown by countless Google search examples:
,
Main heading
It is best to assign a unique, document-specific title for each PDF (and it is also possible to set the PDF so that it shows the title in the title bar when opened locally). If you don't plan to assign document-specific titles, having no title at all is much better than a default title.
They are alive and well in both books, but went astray. You can find them under the bookmark for the heading preceding the Index:

When PDFs are authored with FrameMaker (as is the case here), common reasons for incorrect bookmark order are multiple flows in the same document or disconnected pages.
[Also common to these two guides: the Contents pages don't have any links]
Pages User Guide (3MB) | Keynote 2 User Guide (2.64MB)
| Windows (zip) | Mac (hqx) | |
| Number of pages | 480 | 453 |
| PDF file size | 3.8MB | 7.77MB |
| Front matter page numbers | Arabic numbers | Roman numerals |
| Body text font, size | Arial TrueType, 10.98 pt | Helvetica PostScript, 11 pt |
| Text display quality (same Windows system, 70% zoom) |
|
|
| Title page graphics | Full page | Cropped |
| Number of Bookmarks | 0 | 0 |
| Number of Links | 0 | 0 |
| Meaningful PDF title, set to display in the title bar | No | No |
Coloring text in white to make it invisible is a common practice, but it is not recommended with PDFs used on-screen. In PDFs, the white text may be displayed (or even printed) in some circumstances. For instance, Acrobat/Reader supports different color schemes for background/text, through Edit > Preferences > Accessibility, even when the PDF is displayed through a web browser -- and when the background color is not white, white text shows up. In addition, "hidden" white text behaves like any other text: it can be selected inadvertently when copying, it is located by the Search function, and it will be spoken by Acrobat/Reader screen reader (View > Read Out Loud).
Examples below are from Adobe's PostScript Language Reference (PDF, 7.4MB) [same problems present in different versions of PDF Reference]:


Notes:
Multi-Edit Software's web site announces: "The New Multi-Edit v9.10 Manual is Online! ... Complete with all the new content ... Please note: This is an online version and the graphics have been compressed for easier downloading...".
But this PDF (289 pages, 2.9MB) has zero links, zero bookmarks. And many potential users are likely to have zero tolerance for such an online manual.

Arbortext's Technical Support Guide (26-page PDF), "created and published using Arbortext Software", has a Contents page with 32 entries which use 226 links -- that's an average of 7 links per entry. The links point to the corresponding pages (and not to the specific location of the heading on a page, which is less convenient if the heading happens to be placed at the bottom part of the page), but otherwise they are all valid.
The PDF has bookmarks, but does not specify these to be displayed
initially. Bookmarks for chapters are present as numbers only,
with no text; bookmarks for the headings include the paragraph
number together with text (in the case of 3- and 4-level headings,
numbering often causes heading text not to be displayed in full
in the bookmark pane).
The default setting in many FrameMaker templates and documents is to create Acrobat articles from text flows. Even when implemented properly, the presence of articles is often problematic, since many PDF users are not familiar with this feature and do not know how to use it efficiently. They may try to click a link, but activate the article instead; they are not taken to the expected location, and the view/zoom changes as a result of the article activation (hiding buttons or navigational items, if present).
A demonstration of another aspect is provided by the PDF file of FrameMaker 7.1 Reviewer's Guide. When providing an article thread in a multi-column layout, it is essential that the article's beads match individual columns, and not the page. Having three columns treated as a single bead is counter-productive. As the article is activated, the three columns are displayed. Additional clicks scroll the three columns as one unit and then continue to display the next three columns on the following page, rather than the top of the next column in the current page.
Note: same applies to the "User Guide Supplement" PDF provided with FM7.2
[In Acrobat/Reader 8.0, Edit > Preferences, General, Basic Tools, you may turn off the "Make Hand tool read articles" option]
These and several other cross-references in FM7.0/FM7.1 online help point to page numbers, which are of course irrelevant in an online help format. Moreover, the page numbers do not even correlate with the relevant information in the FM7 printed user guide.
[FM7.2 follow-up: the help system for 7.0 was used "as
is" in 7.1 and 7.2, with this and other problems.
Thus, according to the version information file, the FM7.2 help
was compiled on March 14, 2002 -- i.e. three and a half years
before the actual product was shipped.]

Have you ever tried using the FrameMaker Character Sets (Windows) document
(stored in FrameMaker's OnlineManuals folder) to look up the
access code for Symbol and Dingbats characters? Don't bother,
as the information is not always correct.
The mistakes have not been fixed since this document was published
in 2000 as part of FrameMaker 6.0. Will the PDF in the forthcoming
FrameMaker 7.1 release be more dependable?
[follow-up: same in FM7.1 and in FM7.2]
When searching for PDFs on the web, you may be surprised to
see results displayed with titles that have random-looking letters,
such as oooooo, aaaaa, ggggg, rrrrr, or even !#"%$.
A common reason for these strange-looking titles is missing DocInfo/metadata fields. In their absence, web indexing mechanisms resort to displaying the title using what seems to be the beginning of the content -- which may have duplicate text strings (sometimes used by applications/drivers to create a "bolded" version of the text), text-based shadow effects, decorative dingbats, special symbols or form field elements. Some font-related problems may also cause the underlying text to be very different from the text displayed in the PDF page.
Advanced typographical features available with OpenType fonts
(including ligatures, small caps, oldstyle figures) help produce
superb type (print and display), but may cause problems when
the PDF is used online. Acrobat does not recognize some of the
resulting characters (even when it's a Tagged PDF), so that although
you see the words on the screen, the internal text representation
is garbled -- disrupting the Find/Search and Copy/Paste functions.
In the paragraph shown above (page 2), you won't be able
to find the words file, SFNT, This
or suffix either.
[a new version of this document, produced with
InDesign CS, fixes many -- but not all -- text-related problems]
Search Google for information on "Word to FrameMaker document conversion" and you'll find... PDFs of user guides for gateways and modems. Confused? See Dec. 2002's Hmmm... for hints on PDF document titles.

You can do so (sort of) in the Java Coding Style Guide (PDF:
256K) from Sun Microsystems' white paper library.

Acrobat 6 Core API Reference was made available (PDF: 12.1 MB)
late May. Still labeled as "draft", it does not have
even a draft table of contents for its 3357 pages.
Update (11/03): The final version, dated November 25,
2003, does not have a table of contents either.
Note: The Acrobat 5 Core API Reference PDF was briefly reviewed, among other PDF files included in the Acrobat SDK documentation set, in PDF Best Practices #3: Links Can Guide the Way (12/2001): "...You would expect that CoreAPIReference.pdf, being a 2750 page PDF, would have a detailed and fully-linked table of contents, making it easier to locate items or navigate in a document of this scope. The Table of Contents is slightly longer than one page (!) it has 19 items in total. And to make it even more absurd, these items are not even linked..."
Sun MicroSystems' Product Documentation site has quite a few PDFs, authored with FrameMaker. The link area in tables of contents is inconsistent, sometimes containing only one word or one character, and does not include the page number. Below are sample link areas as displayed when Acrobat's Link tool is selected (Acrobat Reader users need to "hunt" for links by moving the Hand tool and watch for the change of cursor).

Sun StorEdge Enterprise Backup Software 7.0 Release Notes
(816-7890-10.pdf)

Sun StorEdge Enterprise Backup Software 7.0 Roadmap
(816-7891-10.pdf) -- not even one link or
bookmark!


Sun SAM-FS and Sun SAM-QFS Storage and Archive
Management Guide (816-2544-10.pdf)
To make it easier for readers to provide feedback, the e-mail
address could have been a link, opening an e-mail message with
recipient, subject line (with the correct document number) and
some message text. A bookmark would serve this function even
better, as it is more visible (example of e-mail links/bookmarks).
[a message was sent to this e-mail address on May 27, 2003, still
waiting for a reply]
The QuickTime 6 API Reference -- which "documents every function, struct, data type, and constant in QuickTime" -- is a whopping 36M PDF file! With very little effort (and without even having access to the source files), the file size can be reduced to a 14M PDF without any loss of functionality.
Over 50% of the 36M PDF is related to superfluous named destinations. But this reference-oriented PDF isn't that generous with bookmarks. There are only 25 bookmarks for 2200 pages of alphabetic function listing. Want to locate information on the MusicSetPartInstrumentNumberInterruptSafe function? Well... click the M bookmark, which takes you to page 832. The item you need is only 248 pages away (page 1080) -- use the page down key or scroll to get there. Sure, you can also use Acrobat's slow Find function, if you are patient enough and can type the function name without any typos. Considering using the Table of Contents to quickly get to the item you need? Don't bother, as it is not any more detailed than the bookmarks.

Adobe Acrobat 5.0 at Work: Engineering How To
Guide (3MB / 76 pages) discusses how
to "create reliable Adobe PDF files for efficient exchange
of engineering documents" and more. The introduction states:
"Acrobat converts files to Adobe Portable Document Format
(PDF), a universal file format that preserves all the fonts,
formatting, graphics, and color of source documents, regardless
of the application and platform used to create them." But
not regardless of appropriate ingredients and parameters, as
this PDF file demonstrates. When fonts are not embedded, Acrobat
resorts to font substitution with variable results. When excessive
downsampling is used, images and bitmapped text may lack significant
details.

The documentation download page for Enfocus Instant PDF has links to 16 different PDFs. The document developers certainly made their life simple - they used the same PDF document title ("Enfocus End User Documentation") for all PDFs, regardless of its relevance to the content or to the document language. This generic title is also used in PDFs for other Enfocus products.

The PDF DocInfo Title field should be highly specific, as it is

Qualcomm's documentation download center contains PDFs of four identical versions of the GSP-2800/2900 User Guide in these languages. Three weigh approximately 1.2MB, one weighs 4MB.
French may be tough for some people to learn, but having examined the PDFs using Acrobat's Tools > PDF Consultant > Audit Space Usage function, it seems to me that in this case it is the graphics to blame, not the language. The French version was distilled using drastically-different job options. No one seems to have questioned the significant file size difference, not even the web master who diligently indicated the file size for download purposes.
Q: Which markets require document usability? A: Apparently Spanish-speaking countries.
The Spanish version of the above document is the only PDF with basic interactivity -- bookmarks, linked table of contents and index (but unfortunately 7 links in the 1-page index are invalid).
Moral: Always thoroughly test PDFs before distribution!
Scanned PDFs are associated with lower display/print quality, excessive file sizes, non-functional Find/Search functions (unless an OCR layer is added, and even then, searchability may be limited).
Why would someone scan a high-quality, text-based PDF into a bitmap-based PDF? In the case of the freely-available PDF of Acrobat Shortcuts, it seems that the reason was the unethical deletion of the MicroType logo and line (including the URL, where an updated version may be found).
This is MicroType's Acrobat Shortcuts PDF, discovered at the
U.S. Department of the Interior's Minerals Management Service
web site:

This is the original version (free and updated) of the Acrobat
Shortcuts PDF (116K), still available
at the Resources section:

Adobe FrameMaker 7.0 Solutions Guide (PDF: 7.65MB / 123 pages), page 9-3: "FrameMaker... maximum PDF publishing integration... automatic bookmark generation, automatic production of hyperlinked tables of contents", yet these were created manually in that PDF (authored with FM7.0).
When selecting the Link tool while in the TOC pages, one sees uneven/overlapping links (not created automatically by FrameMaker):

The random extra spaces in some bookmarks indicate manual creation (New Bookmark when text is selected with Text Selection tool). The extra spaces are also an indication of text-related problems in the corresponding pages, and will affect Find/Search or copy/paste operations.

[September 2000] Is This a Useful PDF Reference Point? Adobes
own PDFs offer no benchmark for users.