Ubuntu 11.10 VM including OpenJDK Build Image

[Wade Chandler]

On 02/24/2012 09:29 AM, Andrew Haley wrote:
> On 02/24/2012 01:58 PM, Wade Chandler wrote:
>> On 02/24/2012 04:20 AM, Andrew Haley wrote:
>>> On 02/24/2012 02:21 AM, Wade Chandler wrote:
>>>> That forces Linux distros to distribute OpenJDK only and users to
>>>> get Oracle JDK 7,8,etc on their own if there is to be such a thing
>>>> and obviously depending on its price.
>>> Well, OK.  So what?  Why would anybody care?  OpenJDK is just fine.
>> This is all my opinion of course, we have gotten down into this
>> conversation, but all the points here are related to exactly the reasons
>> we have been talking about. So, here is the rest of my thought process
>> on this topic, and will just leave it at that.
>>
>> To answer "So what? Why would anybody care?" Really?
>>
>> What is OpenJDK? Every Linux distros OpenJDK is what? Testing is
>> what?  Users download what? Today, many applications have issues
>> running on OpenJDK because of differences. If OpenJDK is the defacto
>> Java, and it has issues compared to the previous defacto Java, and
>> TCKs are not run on distributions for popular operating systems
>> easily grabbed from the site, and there are no binary downloads
>> available, or at least this is perceived because of having to find
>> each and everyone in one of hundreds of projects, then what is it,
>> and how can many application developers use it, and without those
>> end user developers, how does it continue to be as strong as it is
>> today?
> We've got to solve the true problem, not patch around it.  If the
> OpenJDK on some of the distros is defective or untested, we are not
> going to solve that problem by providing a binary download that hasn't
> been tested on that distro.  That just takes us back to the dreadful
> state of affairs before all the distros packaged OpenJDK.
>
>> For me the binary distribution is a big part of this. OpenJDK may be
>> more important than NetBeans or Eclipse, but only because there are
>> more applications using it.
> That's a reasonable operational definition of "important".
>
>> Without good IDEs there won't be many developers writing
>> applications for it. The new comers will be doing what all of us did
>> when we were younger as it relates to Java, but with different
>> technologies; using the free IDEs for C/C++, Python, PHP (which I
>> also do today, but still a valid point). If it doesn't remain as
>> easily accessible as Java has in the past I am of the opinion it
>> will not remain on top, and my reason for these points on easily
>> downloadable binaries.
> All of these are easily downloadable on all the distros.  That's what
> distros do very well.  I remember the bad old days of free software
> where you really did have to grab a compiler from here, developer
> tools from there, and so on.  It was horrible.
>
>> For instance, upcoming versions EA releases or simply validating
>> whether OpenJDK runs applications running on the Oracle build or
>> older versions of the JVM.
> Fair point.  That is a valid technical reason for keeping a bunch of
> old Java binaries around.  It would be a valuable resource for
> developers.  It would be a Bad Thing, to say the least, if end users
> ended up using these downloads.
>
>> Should folks wanting to help test those things out have to build the
>> system? Seems that loses a lot of folks who could help track down
>> bugs and make it better using their real world applications.
>> They "could" still do that, but it comes down to time. In most cases
>> those folks are paid to work on business/domain logic. Testing and
>> getting ready for new platform releases is part of that. Building the
>> JDK probably not, and that isn't their fault. Some are paid to work on
>> OpenJDK, and for obvious reasons, it is the basis of large vendor
>> platforms. Both types are extremely important.
>>
>> So, I'm not knocking it. I want it to succeed. I think there are certain
>> things which can make that more probable.
> Fair enough.  But encouraging people do use binaries that have been
> put on a site that may or may not have been tested and may or may not
> have any kind of update mechanism is *dangerous*.  I can't stress
> this enough.
>
>> But, too, that all depends on what the exact goals of OpenJDK
>> are. If some of these things are not goals of OpenJDK, then that is
>> fine too. If the goal of OpenJDK isn't to produce any binary result,
> Of course it is!  But you don't get a binary by just building it, or
> at least a binary that a reasonable person would trust.
>
>> but to instead be a basis for the involved vendors brands of JVMs,
>> then I believe that should be stated some where on the front page as
>> many people use Java and have dedicated their time to it.
> You're saying it as though OpenJDK not providing binaries is a bad
> thing.  It's not.  OpenJDK is, and has always been, a source
> distribution, for the reasons that I have tried -- apparently
> unsuccessfully -- to explain.
>
> It would be fantastic if someone could put in place a mechanism to
> build and update properly tested OpenJDK binaries for all operating
> systems.  But it is not just a matter of doing a build and putting it
> on a web site somewhere: with distribution comes responsibility.
>
> I admit that this is a free-software-centered viewpoint.  I apologize,
> but that's the only viewpoint I have.
>
> Andrew.
No worries Andrew. It is a valid viewpoint you have. I just wanted to 
stress my points, and you have communicated yours very well. On some 
points I very much agree with you. Others I think it isn't as black or 
white, but I think that it is OK to disagree on some things. Thank you 
very much for the discussion.

Wade

-- 

=================
Wade Chandler

Software Engineer and Consultant
NetBeans Contributor
NetBeans Dream Team Member

wadechandler.com
netbeans.org