[jbossws-dev] Re: Web Service Stack for EAP 5

[Mark Little]

On 19 Mar 2008, at 18:42, Burr Sutter wrote:
> I'll have to say that that I don't fully agree with "good enough". :-)

Don't believe all of the hype around WS-* ;-) Having been involved  
with it since 1999 and co-authored more of the specs/standards than I  
care to remember, I'm fairly happy to say that with some confidence.

> However, it is very hard for me to specifically state that a  
> particular feature/item "must" be addressed in a particular time  
> frame.  Here is why:
> - We don't have a formal way of surveying our userbase to see what  
> would really be interesting & compelling for them.  We have to gave  
> into the proverbial crystal ball and try to back up our  
> prognostications with anecdotal evidence from personal visits with  
> customers, forum postings and support cases (so I added Darran on  
> the CC, do double check my list of priorities below).  Certainly the  
> jira voting system has failed us in this area.  So we lack real  
> compelling and actionable data so we base things on intuition.
> - Many of the people who do ask for a specific WS-* fall in the  
> "tirekicker" category.  That is a US expression for someone who is  
> basically a professional shopper, not a buyer (aka adopter/user),  
> they love to look but they have no motivation to make a decision.   
> Corporate America's IT shops are full of these kinds of people.   
> There are people who have full-time jobs in product review,  
> standards setting and general vendor abuse. :-)  These folks are  
> normally measuring the "value" for technology based on its buzzword  
> compliance level, much like a Gartner or Forrester would as well.     
> I'm not saying they are bad people but their job is to put often to  
> keep particular technologies out of their data centers, they love to  
> say "IBM is our standard, JBoss isn't on the list, unless you can  
> prove X, Y & Z to me".  These folks are non-coders and won't even be  
> involved in the real deliverables. - Some "prospects" represent ISVs  
> who often love the "best of breed" approach.  They will self- 
> assemble their own "platform" from various OSS projects.  Obviously,  
> this is not a "sweet spot" for us since we want people to use  
> specific components in a specific platform so we can properly  
> support & patch their platform. So I try to remove these  
> recollections from how I prioritize things in my own head.
> - I have the perception that the "big vendors" have dedicated teams  
> of "standards" folks who are pushing BS in the various committees to  
> simply stay ahead of the open source commoditization of their  
> techs.  Let's face it.  The big boys have lost billions to open  
> source (JBoss, Tomcat, Hibernate and others), I'm fairly certain  
> that they see "standards" as a competitive weapon and their sales  
> staffs have been instructed to educate the userbase on this fact  
> that open source can't keep up in standards & tools.

They do. Oracle get a lot of business on this fact. But we are  
standards compliant. No one is compliant with all of WS-*. That is not  
the point. We just need to pick the core. What we have is the core.  
That is what my comment about "good enough" referred to.

>
>
> With all of that said, WS-RM & WS-Notification are asked for the  
> most (beyond SOAP, WSDL, JAX-WS, what I call the basics).  There is  
> a belief in our userbase that they do wish to take advantage of  
> these features even if they don't yet know how.

Well we need to educate them that WS-N is dead for start!

>   I don't recall running into a single JBoss shop (outside of ISVs)  
> who specifically moved to CXF or Metro to get this feature.   
> However, I have run into several who continue to use Axis on JBoss  
> for various reasons (mostly around legacy MSFT interop or simple  
> comfort factor).
>
> Honestly the most actionable priorities that I feel "strongly" about  
> are (granted, not are all on this particular team), listed in  
> perceived priority order:
> - Dramatically better integration into the ESB
> - JBDS Tooling needs to have support for JBossWS, not just Axis,  
> customers are confused by this
> - JON needs to have support for JBossWS: take what is at http://localhost:8080/jbossws 
>  and make sure it is available in JON, perhaps it is there but I've  
> never seen it.
> - More proof of interop with .NET.  At this point, I just tell  
> people that if they find it is broken, get a support subscription  
> and enter it as a high priority bug, we will fix it.

Agreed and that's what I said before.

>  I remember Kevin W from MSFT worked his butt off at JBossWorld  
> Vegas to get helloworld version of WS-Security & MTOM working with  
> a .NET client to JBoss backend.  It wasn't well documented, no  
> examples, lots of pain on his part and he would be considered a  
> very, very advanced developer, well beyond the skills of your  
> typical enterprise IT developer (our typical customerbase)

Yes, I can understand that. We do not participate in interop enough.  
We also do not make our endpoints public, which is something I've been  
pushing for since 2005.

>
> - Security: there is a lot related to security that I've not had a  
> chance to think through.  It relates to JAAS, SAML, integration with  
> Red Hat's "certificate server".  This actually comes up often and I  
> dance enough to skip the answers. :-)  However, there is enough  
> questioning in this area to suggest the userbase wants some nice  
> examples/docs and education on the "best practices".  Clearly I have  
> no clue how to implement all of this but I've not dug through the  
> docs, wikis & forums enough to see if someone has described how to  
> tackle all of my security (encryption, authentication,  
> authorization) needs.  I've never seen this in JON and that is where  
> it eventually needs to get to as security is not managed by the  
> developers but by the administrators. - REST
> - SOA Software & Amberpoint supported
>
> So, I ask all of you this.  Do the bullets above also sound like  
> priorities in your own minds, therefore, we/the team, need to spend  
> a few more cycles really nailing down these requirements, designs  
> and roadmap which release/timeframe they'll be "proven" in?  Is  
> there anything that I left off that you happen to know is also a  
> critical priority (that isn't just a bug fix).

Priorities for what? 2008? 2009?

Mark.

>
>
>
> Mark Little wrote:
>> Understood and I think what you've got at the moment is really good  
>> enough. It's the core capabilities. Let's just make sure it's  
>> interoperable and fast, and job done :-)
>>
>> Mark.
>>
>>
>> On 19 Mar 2008, at 13:11, Heiko Braun wrote:
>>
>>>
>>> Agreed. I am just saying instead looking at what the others
>>> have on their managers checklist, we should first come up with
>>> actual requirements.
>>>
>>> If we are talking about EAP 5, just write down what the stack really
>>> needs to provide and the WS team can decide on the best way to  
>>> achieve
>>> this. That's at least my opinion.
>>>
>>>
>>> /Heiko
>>>
>>> On Wed, 2008-03-19 at 12:57 +0000, Mark Little wrote:
>>>> But it's not the way to go in the future.
>>>
>>
>> ----
>> Mark Little
>> mlittle at redhat.com
>>
>> JBoss, a Division of Red Hat
>> Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod  
>> Street, Windsor, Berkshire, SI4 1TE, United Kingdom.
>> Registered in UK and Wales under Company Registration No. 3798903  
>> Directors: Michael Cunningham (USA), Charlie Peters (USA), Matt  
>> Parsons (USA) and Brendan Lane (Ireland).
>>
>

----
Mark Little
mlittle at redhat.com

JBoss, a Division of Red Hat
Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod  
Street, Windsor, Berkshire, SI4 1TE, United Kingdom.
Registered in UK and Wales under Company Registration No. 3798903  
Directors: Michael Cunningham (USA), Charlie Peters (USA), Matt  
Parsons (USA) and Brendan Lane (Ireland).