Re: What Java did right (was Re: Y Store now C++)

["Anton van Straaten" <address@hidden>]

Sundar Narasimhan wrote:
> There are problems that I had in 94-95 that were not solved by
> languages that were available then -- as there are today. Some of them
> I could paper over with JNI calling fortran or C. And many of the
> languages that people discuss on this list weren't around in the time
> frame I'm talking about. I am saying I've talked to people who made
> decisions to adopt Java in their work, and that they seemed
> intelligent and not the kind to be swayed by marketing hype alone..

I'd put myself in that category.  I downloaded a Java beta around the end of
'95 (I think).  Applets were obviously just fluff, but the language seemed
to have a lot of potential for database application development.  I tried it
out for some minor projects in '96.  Some of my clients began using it in
'97, for server-side development.

The early emphasis on client-side GUI development mystified me - Java seemed
too slow, too klunky (non-native widgets).  Corel's attempt to develop an
office suite in Java seemed doomed from the start, and of course it was.
But none of that had any bearing on the potential for server-side
development.  It took the tech press years to pick up on this idea, though -
I saw a big disconnect between why tech people were actually interested in
Java, and what the press was writing about Java.

To put the choice of Java in some perspective: in 1995, many of my customers
were running MS Windows 3.1, i.e. MS-DOS in all its primitive 16-bit,
souped-up with a GUI that had a retrofitted 32-bit API - not the most
capable platform.  Some of these customers had PC-only environments with
Novell or Windows NT file servers.  Languages commonly used were C, C++, VB,
Pascal, and a slew of low-end "4GL"-type languages, including dBASE and
derivatives, and things like Gupta's SQL Windows (a GUI development system
for two-tier client/server applications).

In that world, Java was a Great Leap Forward.  There weren't a large number
of other options.  Three that come to mind are Borland's Delphi, Python and
Smalltalk.  I won't try to enumerate all the reasons that we didn't choose
to adopt those languages, but it would be all the usual reasons, with an
important one being perception of the language's future.  On that count,
Delphi and Smalltalk were virtual non-starters; and the jury was still out
on Python, at the time.

This is an area where Sun's marketing, and financial commitment, made a
difference: it was possible to predict Java's incipient mainstreamhood, you
could watch important progress happening over a period of months, you could
see it gathering momentum.  The portability was a potential advantage.
Microsoft initially gave it a stamp of approval by releasing its own version
of Java.  It very quickly reached a point where it appeared a "safe" choice.
In hindsight, that perception proved correct.

The marketing emphasis on browser applets might be an interesting lesson in
language adoption, though.  It acted as above-the-line advertising which
raised awareness of the Java brand amongst people who wouldn't normally hear
about languages, including CEOs and CFOs.  I remember one CEO asking about
"HotJava", and you could see he just really liked the name - he was
disappointed to find out that HotJava wasn't the name of the language.

It's a little like Intel's advertising which claims that its CPUs improve
your Internet experience - as though I need a GHz P4 to process cable modem
bandwidths.  The specific claim doesn't really matter - it's the brand
awareness that's important.  If a CEO or CFO has heard of something as
esoteric as a computer language or CPU, and if it's coming from a big
company with a well-known name, that's an easy sell.

One might think it shouldn't matter whether a CEO/CFO has heard of a
language, but CTOs, IT managers etc. aren't all so secure in their jobs that
they don't get some benefit from their boss recognizing, and having warm
fuzzy feelings about, their choice of key technology.

Anton