PDF Document Management Software, Services & Support
| Server | Desktop | Services | Support | Why Us? | About Us |
PDF Readers - Review Notes
Size Matters, or Does It?
To put the question of "bloatware" in context, we recorded typical "bloat" metrics for all applications, specifically:
- The installer size for each application as downloaded.
- The installed size of each application (ie, the total size of the files installed into the /Program Files folder)
- The RAM consumed by each application (each app had the same PDF open for this test).
Look over these numbers, then compare with application performance.
Yyou be the judge. Does size really matter?
This 1018 page PDF (one version of the US House's Health Care Reform legislation) was used for the main speed tests. Default settings did not include thumbnails (since they add a lot of time). The timekeeping was manual; your mileage may vary.
The cold start speed is the average of three trials following a system restart. This speed will typically be encountered only the first time an application is opened after a system restart.
The warm start speed is the average of nine trials (actually, ten trials with the outlier eliminated before averaging). This approximates the performance encountered when re-opening the application between restarts.
For the open file speed test, the application is already running, so we can focus on the time required to simply open the PDF. This figure is also the average of nine trials (after one outlier result was rejected).
The "open big file" test was conducted on a press-ready PDF, in this case, a 525 MB, 244 page file created in Quark. The file includes tons of fonts, vectors, images, you name it. This figure is the average of four trials, after one outlier was rejected. For obvious reasons, I'm NOT posting the file - you'll just have to trust me!
Open Options
Not all viewers respect every possible setting in PDF files with respect to the display of bookmarks, thumbnails, magnification, initial page number, toolbar display and other so-called "open file options" that may be set by a document author.
A PDF viewer must be able to display the document as the author intended. Bookmarks and thumbnails are vital usability features for longer documents, and other file option options (2-up display, or initial zoom level, for example) are important as well. These settings aren't new to PDF; this is a fairly basic area of functionality, and all prime-time PDF readers should get this right.
We tested each Reader in this survey to find out if they respected five possible non-default settings. Two Readers showed one error apiece: Tracker's Viewer failed to display the bookmarks and Nitro's Reader failed to hide its toolbar. We will probably perform an exhaustive test of the Open Options features in the next survey, so developers are hereby warned!
Signed and Certified PDFs
Using each Reader in turn, we attempted to open this PDF file (the US Budget for 2011, 7.7 MB), a GPO-certified (ie, digitally signed and secured) PDF. We then noted the differences in behavior.
“Full Support” means that the reader notified the user of a digital signature, accurately reported its status and provided a means to validate the signature (or certificate).
“No verification” means that the reader was aware the PDF was signed, knew the signature was valid or invalid, but was unable to verify an otherwise verifiable signature.
“Error(s)” means that the reader stated that a valid and verifiable (by Adobe Reader) certificate was nonetheless invalid and/or reported an invalid certificate as valid. This is most certainly not the desired result.
Support for PDF/A
The archive subset of PDF is ISO 19005, better known as PDF/A, is the standard for long-term preservation of electronic documents. Published in 2005, PDF/A is being adopted by governments, courts and businesses around the world as a key part of their information retention strategy.
PDF/A includes "conforming viewer" requirements, which Adobe presently over-interprets to mean "read only". Quibbles about how to implement "PDF/A mode" are one thing, but only one application (Adobe's) appeared to even notice when a PDF/A-flagged file is opened!
Software developers should be sure to follow best practice when implementing any International Standard, and PDF/A is no exception. This article offers guidance on developing software to work with PDF/A files.
At this point, it's fairly inexcusable for desktop PDF readers to simply ignore the PDF/A flag, but all the Adobe-alternatives still do precisely that. C'mon, developers; you can at least read the XMP and show a flag, even if you aren't validating for PDF/A compliance!
Annotations & Stamping
The test file is a simple PDF that includes one example of each type of standard PDF annotation.
While there are many powerful annotation features in Acrobat and Adobe Reader, a competing PDF Reader, minimally, must be able to properly display standard annotations added by other software.
To gain an advantage over the 800 pound gorilla, Adobe's competition has spent a lot of time and effort on providing annotation creation features in their free Readers. In Reader X, Adobe strikes back, for the first time adding Notes and Highlighter tools to Reader without requiring Reader Extensions.
Inexplicably, however, Adobe chose to drop some of the standard functionality for Notes created in Reader X; you can't reply to a Note in Reader unless (you guessed it) the PDF is Reader Extensions-enabled! This earns Adobe an "Incomplete".
Only two applications in the review, Tracker's Viewer and Nitro's Reader, offered any type of stamping - the addition of text or images directly to the PDF page. This is a powerful feature, and we expect to see it cleverly implemented in future versions of these increasingly assertive "Readers".
* - Only two applications, Adobe's Reader X (as noted above) and Nuance's Reader, are unable to reply to an annotation. On the other hand, Nuance's Reader was the only non-Adobe application to successfully play an audio annotation (my personal favorite).
Forms
We attempted to open a few of our customer's forms with each viewer and noted the results.
“Simple” forms - standard Acrobat forms (Acroforms) including only basic calculations and validation.
“Complex” forms - standard Acrobat forms including field-properties formatting and other advanced Acrobat JavaScript methods and techniques.
Both static and dynamic XFA forms were tested. Adobe's Reader handled the XFA with aplomb (as expected). Nuance's Reader did a credible job on XFA forms of both types. Tracker's been working on XFA as well; while we saw problems we also saw a lot of progress.
** - Of the non-Adobe PDF Readers, Foxit showed promise in terms of improved support for advanced PDF forms and Acrobat JavaScript functionality. BUT - we still (as in the previous review of this software) saw some errors when rendering (displaying) forms including blacked-out form-fields.
Reuse (exporting text & accessibility)
There's something a lot of people want "out of" PDF files, and that thing is text. PDF are internally complex, and most lack the structure (known as "tags") necessary for reliable results when exporting text. Worse still, few Readers even bother with tags.
For this test, we extract text from a PDF using two different methods. In one case, we export the entire document; in the other, we select some text on a page with the mouse, copy, then paste into Notepad. The document is correctly tagged, but the content stream's reading order differs from the tag order, allowing us to determine how each Reader goes about generating usable text.
Reads Tags: Users who rely on assistive technology require properly tagged files in order to read. Only PDF Readers capable of reading PDF tags can comply with Section 508 regulations regarding accessibility.
Export Text: Only Adobe Reader uses tags when exporting text. Two competing Readers (Foxit and Nitro) ignored the tags but include software that interprets the page, attempting, in effect, to tag it "on the fly". Results are guaranteed to vary.
Copy & Paste: None of the applications use tags to support copy and paste; and Adobe and Nitro add some some interpretation to the raw "reading order". Page interpretation will give better results than blindly following the content-stream in many cases, but both are poor substitutes for tags.
*** - The Nuance Reader does not include an option to export document text. It requires the user to upload their file to Nuance's servers for processing, which involves providing an email address. I dutifully tried this, only to learn that their servers "couldn't convert" the same file the other free Readers were converting right there on my desktop - that's a #fail!
Leave a comment.
