r



6 Important Factors when
Choosing a PDF Library
ADAM PEZ



Overview
You plan to embed PDF functionality into an
application. But before you dive into the project,
you must decide: do you go with a more expensive commercial PDF SDK — or a lower-cost
alternative such as an open-source library or
an open-source wrapper?
There are non-trivial costs to switching later.
Developers have to re-learn the new library,
re-adjust the backend, customize the UI to match
what users are accustomed to, as well as migrate
documents, form data, annotations, and more.

This dissatisfaction implies that picking the right
PDF SDK is a lot harder than it seems. And to
help you avoid the same mistakes as past
implementations, we’ve written this article.
(We also recently surveyed 57 unique
organizations that switched from PDF.js to a
commercial SDK. Read our comprehensive guide
to PDF.js to learn more.)

According to market research conducted by Stax
Inc., the average Net Promoter Score (NPS) for
the top five PDF SDK vendors is 35%. And 70%
of customers express interest in switching
despite the high costs.

2



Restrictions on Features and Platforms
A first mistake organizations make when
selecting a PDF library for the first time is to
assume fixed feature requirements. But these
are likely to evolve.
Users start to ask for more functionality as they
grow dependent upon a PDF SDK. An organization will then have to consider saying no to user
feature requests; building time-intensive and
challenging customizations on top; or integrating
additional libraries and thus adding more
complexity, maintenance overhead, and risk.

Additionally, a library may work fine initially on
the main platforms preferred by your users. But
later if you wish to expand, the library does not
support the platforms you want — or the APIs are
inconsistent, with different classes and methods
across platforms making it so your engineers
have to start from scratch.
To avoid this hidden cost, go with an SDK
with a broad feature-set across multiple
platforms, providing you the flexibility to
grow down the road.

Maybe big companies can absorb the costs of maintaining
three-to-four different relationships with different vendors,
each with a different code base, different roadmaps, and
different problems. I’m not saying it isn’t possible.
— Kalsefer Co-Founder and CEO, Avshi Segev

3



Unanticipated Difficulty Adding Features
Another mistake is where organizations select a
basic library to save money with the assumption
that they can build anything needed on top.
But building in an unfamiliar domain can easily
lead to unknown challenges, high expenses, and
reduced speed to market. And PDF is unusually
complex as high-profile teams attest — including
those of Slack, Dropbox, and Linkedin.

involves a steep learning curve not subject to
economies of scale. Throwing more devs into the
equation will not shrink the ramp-up time for the
first developer.
Additionally, custom features will have to be
supported and maintained long-term, creating
an additional ongoing opportunity cost: committed resources will be less-available to work on
other projects.

Your devs are not necessarily PDF experts, and
attempting challenging PDF features in-house

PDFs are an incredibly complex file format; this is especially so
given that a PDF can be generated a hundred different ways,
all of which a renderer needs to handle gracefully.
— Developer, LinkedIn

PDFs are complex documents
— structured into different
layers of information, data,
and objects, and containing
different languages, images,
and graphics.

PDF is an incredibly complex
file format — the specification is more than a thousand
pages long, not including the
extensions and supplements.
— Developer, Dropbox

— Developer, Slack

4



Organizations that we’ve spoken to have found
the most challenging features are those that
require engaging PDF at a low level, where
objects are defined in PDF byte code —
with unique byte offsets for different objects,
making it difficult for devs unfamiliar with PDF’s
inner workings to parse and manage these
objects correctly.
Challenging PDF functionality includes
• Managing PDF annotations from multiple
users (e.g., synchronization and versioning
• PDF generation (creating PDFs from scratch
or from other documents)
• Page manipulation (add, merge, or remove)
• Layers (via Optional Content Groups)
• Color management features (e.g., ink-color
separations, overprint, etc.)

While it is certainly possible to build the above
in-house, PDF features can consume a shocking
amount of time. And you eventually may have to
decide whether to continue — or whether to bite
the bullet and abandon months or years of work
for an alternative that can meet your requirements cost-effectively.
To avoid this type of hidden cost, you will want to
carefully consider the capacity of your existing
development team should you decide to build,
maintain, and support custom PDF features inhouse as these features often prove time-intensive and challenging.

...you shouldn’t build anything that’s available off the shelf because it’s not a source of competitive advantage if everybody
else can avail themselves of it. The only scenario where you
should build is if it’s your core technology -- the core source of
your competitive differentiation and competitive advantage.
— Mark Holst-Knudsen, President ThomasNet @
MIT’s 2014 CIO Symposium

5



A lower-quality library also encounters
performance and memory issues, such as large
documents with frustratingly long wait times for
your users as well as complex documents that
crash the viewer. This is often due to the absence
of features such as PDF tiling, parallelization,
and linearization that a more mature PDF SDK
will incorporate.
Some solutions (e.g., image servers) perform
excellently when tested on a small number of
documents and users but then inflict unexpected
hidden costs when scaled up. When hundreds
or thousands of users later view, mark up, comment on, and otherwise interact with (i.e.,scroll,
pan, and zoom) documents, server resource and
network data usage explodes. To maintain your
desired UX, you have to pay higher fees or invest
in more servers.
The following types of documents have much
more demanding rendering requirements:

• CAD-based PDFs such as construction and
engineering drawings with very large and
complex designs.
• Reports, textbooks, and marketing material
using advanced PDF graphics such as shadings, gradients, soft masks, and patterns.
• Geospatial maps with OCG layers that are
switched off by default.
• Pre-press documents which require an SDK
with advanced color management features to
print colors accurately.
• High-speed accurate rendering (especially on
native mobile apps and mobile browsers).
• Context extraction of tables, text, etc. with
document structure (e.g., text read order or
table arrangement) in tact.
To prevent crashes, slowness, and rendering
issues from disrupting your UX, test functionality
with the types of documents your users will work
with. Also test a server-based solution at the
anticipated load and usage.

6



Poor UX: Slow Performance, Crashing,
and Inaccurate Rendering
Another source of hidden costs can be a poor
user experience, especially as users start to
upload more massive and complex documents
that crash or freeze a lower-quality viewer.
Construction Computer Software encountered
these issues with a free PDF viewer add-on to its
flagship estimation software.
As is often the case with a lower-quality library,
PDFs render incorrectly. You then have to wait
on the vendor to respond. But a reseller or a

smaller company with many remote developers
may have difficulty providing the same turnaround time and specialized support and
service as a commercial SDK. If they did not
build the rendering engine themselves, they
may not be able to fix the issue — or fixes may
take a long time — because they have difficulty
finding in the code where the problem originated.
If you go with open-source, you may have to fix
bugs yourself.

If you’re looking for a PDF reader for the first time, you better
make sure it can read 100% of your PDF files. Because if your
client-base starts relying on that PDF reader, exactly what
happened to us, they still want the absolute best quality.”
— Tony Cornwall, Construction Computer Software

7



Low Adoption on a Complex UI
In 2018, AEC-software company PlanGrid
partnered with FMI to survey nearly 600 construction leaders from around the world to
discern why construction and engineering
software succeed or fail. The findings report
“Construction Disconnected” identified a
complex UI and inadequate user training as two
of the top five reasons for why technology fails.
Being able to slim down the interface and tailor
feature-sets to specific user groups is proven to
significantly cut down training costs and improve
user adoption. (See our OEC Graphics success
story to learn more.)

source UI will make it difficult to evaluate how
deeply you can customize, optimize, and add new
tools or annotation types to the UI. Therefore,
your team may build out a proof of concept and
make their plans for future expansion — only
to have to scale back their ambitions or wait on
the vendor to adjust the API. A black box UI will
prove especially problematic if your UI team is
very strict or if you have unique UI requirements
(e.g., accessibility compliance requirements such
as ADA/508).
To avoid this hidden cost, choose a vendor with
an open-source UI or make certain your proof of
concept won’t need to change.

However, a closed-source UI will limit you in what
you can customize, and you may not be able to
fully fix the UX. (And by the time you’ve
discovered this, it may be too late.) A closed-

8



Security Issues

The Bottom Line

When writing PDF features from scratch,
developers may be tempted to take shortcuts
to save time. But these shortcuts cause the
solution to become obsolete quickly as devs run
into the exact security issues a more mature toolkit makes a lot easier to solve.

The best way to avoid hidden costs associated
with the wrong PDF library is to perform due
diligence during your evaluation. To assist you
in this process, we’ve written a blog with several
considerations you can add to your PDF SDK
evaluation checklist.

One recent instance our solution engineers have
noted is where developers use JavaScript-based
submit buttons on forms rather than uploading
and parsing data out of forms — which opens up
the system to phishing and middle-man attacks.
Someone could easily edit the button to have
it send personal information to another server,
and then maliciously re-circulate the form within
your organization or send it to end users.

We hope this article was helpful! If you have any
questions, don’t hesitate to contact us.

Vendor Lock-in
Lastly, consider how your data and documents
will be stored. For example, annotations stored in
a proprietary format, such as Brava! annotations
and some versions of JSON, will not be accessible
to users who want to view their annotations with
other tools such as Adobe Acrobat. Moreover, it
will be challenging to migrate these annotations
later if you wish to switch solutions.
A vendor who manages annotations in the ISO
standard for annotations interchange, XFDF, for
example, will eliminate this hidden cost.

9


Fleepit Digital © 2020