epesh
I'm Joseph Ottinger, editor of TheServerSide.com.

Calendar

««Nov 2009»»
SMTWTFS
1234567
891011121314
15161718192021
22232425262728
2930

My Bookmarks

My Top Tags

                                       

Mailing List

My RSS Feeds








Search Box

 

Apple and Intel means HUGE WIN for Java?

posted Friday, 10 June 2005
I just realised that Apple moving to Intel should mean a huge push for Java, if programmers can manage to think a little bit.

I mean, really: Apple suggests fat binaries, with PPC and Intel code? Yeah, right. Let's take a 2.5MB deliverable and crank that up to a 5MB deliverable. (Yes, I'm oversimplifying a lot.)

Just think: if Apple programmers start using Java, that goes away. The target system - you know, the people who buy the software - just need to install a decent JVM, and the deliverable works. No recompilation. In fact, that deliverable would work on Linux, on Solaris, on Windows... it'd be far easier on the developers and testers (well, slightly less on testers), installers can be standardized...

Of course, this has been the case all along. The issues have been JVM speed more than anything else, and the industry perception that Java is tainted somehow. Well, the JVM speed issue is getting better all the time, and I don't know how to convince people that Java's not somehow evil and corrosive.




1. TronD left...
Friday, 10 June 2005 9:35 am

so what you are suggesting is that Mac developers shall drop the very things that make mac os x a better platform than wintel/linux/unix/etc? we want to be able to use the possibilities present in the platform as good as we can, giving our users the very best experience. i am quite a fan of using Java myself, but that is based on web development or other server-side programming. for a fat client i would very much prefer to have the possibilities at my choice present.

after all, mac os x is mac os x. not wintel. not linux.


2. Joseph Ottinger left...
Friday, 10 June 2005 10:14 am

Dunno - I guess if it's worth your time to compile and test everything twice, for a platform that's going away, it's your choice. The alternative is watching the entire market go a year without buying much Apple hardware, until they switch over. Which would you prefer?


3. Anonymous left...
Friday, 10 June 2005 11:28 am

Except for the simple fact that fat binaries have been around longer than Java. NeXT created fat binaries back in the early 90's when they released NeXTStep for Intel. It allowed NeXT developers to deliver one bundle for NeXTStep regardless of the processor architecture. So, how does this mean a boon for Java? Fat binaries are just as tried and true as Java deliverables. Furthermore, deliverable size means nothing to to customers these days as have DVD readers/writers, fat broadband pipes, and boxen come equiped with hard drives that so large they are nearly impossible to fill by the average consumer. Heck, Adobe Acrobat reader is 30 MB and AOL Instant Messenger is 10 or so MB.

The funny part of the development conversation that started with the Intel transition announcement is how little people understand about the power of NeXTStep which forms the basis of OS X. Gosling and Co. have said on numerous occasions that Objective-C, NeXT and its Foundation libraries were a huge inspiration for Java. It was just too far ahead of its time.


4. Joseph Ottinger left...
Friday, 10 June 2005 11:49 am

Fair enough - except how many of us really run NeXTStep nowadays, fat binaries or no?


5. Anonymous left...
Friday, 10 June 2005 5:15 pm

Actually, anyone running Mac OS X today is running a direct decendent of NeXTSTEP, but that's not the point of my comment. The point is that fat binaries are a well-established technology that predates Java. The requirement of their use to achieve cross-platform compatibility will not drive developers to Java as an alternative. This development will drive developers more towards Cocoa, as it is the equivalent of the JVM in the Mac OS X/NeXTSTEP world. The Cocoa runtime carries the same cross-platform guarentees as the JVM except neutral byte-code.