PDA Forum / General / July 2008

What do you mean? I understand the iPhone is a smart PDA but how can
Google be a PDA program?

In article
news:<7f4151ce-8c79-435b-bf29-6887ee189546@f36g2000hsa.googlegroups.com

It depends on what you mean by 'viable'.

The Java platform is the most common, but is rather limited. Windows
Mobile is becoming more common but is limited to quite high-end phones
(and a few non-phone PDAs), Symbian runs exclusively on high-end
smartphones but is very popular (especially outside the US).

The big problem with J2ME is that it provides a somewhat restricted
Java API and doesn't mandate the provision of interfaces that give
access to phone/PDA features.

The restrictions include lack of floating-point arithmetic and support
for only a very limited set of net APIs and no crypto (which is insane,
as most of the devices that support J2ME are internet-capable phones,
and mobile devices need security more than any other). My impression is
that the people who drew up the J2ME spec expected it to be used to
write games, and not much else.

There is an added problem that some devices make only a limited amount
of memory available to the Java runtime, so there is a limit to the
size of application that can be run. This is particularly crippling if
you have to include Java code in your application to do something that
would be provided by the runtime in J2SE.

J2ME would be much more useful if it had mandated standard APIs to
allow a Java MIDlet to access the device's address/phone book, to use  
ir/Bluetooth communications (if available), to use the device's camera
(if any), etc.. Many manufacturers support proprietary modules that
provide some of these functions, but they aren't standard across
platforms. This severely limits the scope for writing general
applications for J2ME ... at best you will have to write different
versions for different manufacturers' devices.

The J2ME applications that I am aware of are mostly either games or
proprietary applications used within a particular business. There are a
few general utilities, but nothing like the number that sprang up for
the Palm in its heyday. That said: there are an awful lot of cheap
cellphones out there that can run J2ME MIDlets, the market is huge and
potentially rewarding if you can come up with a sufficiently compelling
product.

Symbian development is mostly done in C++, but using a somewhat
symbian-specific dialect. You do get full access the hardware, and
(usually, at least) good supporting libraries that give some access to
the device's software environment. The APIs are not all standardized,
though, and Symbian have allowed their licensees to provide their own
implementation of the top-level GUI, so there isn't even a single
standardized GUI API across devices from different vendors (which
allows vendors to maintain their own look and feel but is a PITA for
developers as you just can't ship a single shrinkwrapped application
for all vendors' Symbian-based phones).

If anything, Windows mobile errs in the opposite direction ... the APIs
are specified by Microsoft so there is a high degree of standardization
across products from different vendors, but there is less scope for a
vendor to do anything interestingly different (at least without
breaking the standard). WM development uses familiar Visual Studio
tools, but the target APIs are subtly (and sometimes not so subtly)
different in the mobile devices -- this means that developing a mobile
version of an existing desktop application is often not as easy as one
would like (sometimes scarcely easier than rewriting it for e.g.
Symbian). The tools are of good quality, though, and if you have Visual
Studio Professional for desktop development you already have the tools
for mobile development thrown in for free (that's an indication of how
much Microsoft wants to encourage developers to develop for their
mobile platform).

As always: if you produce a sufficiently useful/entertaining
application there will be people who will want to use it, and who will
be prepared to pay for the privilege. Whether or not you consider
development for the platform 'viable' depends on how many people will
pay to use it on that platform and how much they will pay -- and that
depends on how good your killer app is.

Cheers,
Daniel.