Pensieri di un lunatico minore

Sometimes, I do wonder whether my interpretation of the English language parallels the commonly-accepted version, and today is just such a time. Earlier, I wrote about some comments James McGovern made, and now Mr. McGovern has responded—unfortunately misrepresenting what I said, and conveniently leaving out bits. Let’s tear it does, as usual.

I took a “cheap shot” at the various “industry analysts,” in a wide-sweeping generalization, to which Mr. McGovern says:

Hmmm. Have you ever heard of a highly respected industry analyst named Stephen O’Grady whom within his blog on numerous occasions has positively blogged on Ruby and no one paid him to do so. Are you wandering alone in the wilderness assuming that absolutely every single industry analyst has no integrity and simply won’t talk about the merits of something unless they are paid? If you spend ten minutes reading the blogs of Redmonk, Elemental Links and others, you would immediately disprove your own notions.

I have heard of Mr. O’Grady, and I’ve had various discussions with some people at other small firms, however the truth is that Gartner dominates every single customer I work with, and the industry as a whole. Perhaps that skews my perception, but every major firm I’ve worked with has a similar “pay for coverage” unwritten requirement. As much as I admire Redmonk, they simply don’t cross the radar of my customers.

Glad that you at least acknowledged that some of the Ruby tools still have a little bit of work to catch up to the world of Java but that isn’t the point. The real statement you should have paid attention to was configuration and release management practices which is more than just Ant and its equivalents.

You’re absolutely right, which is exactly why I spoke about configuration and release management as a process issue, not a technology issue. Technology might assist in some ways, but if you don’t have the right process and methodology, the technology means nothing. Since Mr. McGovern only spoke of the technology, I assumed he was only interested in that aspect. Getting process and methodology right is a much harder problem.

So you don’t think it would be useful for Ruby to have the capability of issuing SNMP traps for enterprise applications? Yes, SNMP can be used to attack applications but also adds value in terms of providing alerting mechanisms based on standards? If you don’t like SNMP, what alerting standard do you prefer? Do you not prefer any form of alerting for an enterprise application?

SNMP for that, I suppose, but so can syslog, which is about 100x easier to parse, transmit and suffers from a lower probability of compromise¹. For example, many of the supposed SNMP messages generated by network gear (such as Cisco) are actually just textual syslog messages stuffed into an SNMP packet. Not very interesting.

In terms of XACML, it can be incorporated into building of application logic but in the Java world, J2EE containers such as BEA WebLogic, JBoss, Websphere, etc are implementing. Do you think they are wrong for building XACML support into the container?

Yes. XACML won’t survive long before something else comes along, and this kind of tight-coupling is only fruitful if your goal is to continually force your customer base to upgrade to newer and “better” releases that track the latest standard that you’ve built in. This is the Microsoft model of software, and it doesn’t survive well under the pressure of delivery.

I would like to understand why the Ruby world would be better off? Wouldn’t the applications built on top of Ruby benefit by using whatever standard is most pervasively used and not just one that is open? [..] Let’s say I agree for the moment that the Ruby community should only do non-Microsoft stuff. Other than for small startups, should the masses of large enterprises not care about interoperability? I hope you will be successful with this approach.

I believe that the Open Source community should do its best to not tie themselves more than they have to to various proprietary approaches. Ruby is a multi-platform environment, and tying ones-self to the Microsoft universe would limit the possibility of deployment on many platforms. Encouraging the development of open standards is in the best interest of the community as a whole. Clients who are top-to-bottom integrated on a single vendor platform are unlikely to be probable targets for Ruby, as they will no doubt be worshiping at the .NET temple.

There are actually several flavors of OS for the mainframe. You have Unix System Services which runs on top of MVS and I am in full agreement that this is a beast that should be avoided. Z/Linux on the other hand has some merit in terms of the community considering embracing. My offer as Mr Enterprisey still stands to the Ruby community for anyone that would like access to a mainframe to get a port going.

As someone who got his start on minicomputers and mainframes, I’m quite aware of the IBM architecture, and even the nuance of differences between the various generations and “compatible” parallels (i.e. Amdahl). Linux is the only viable target, unless you’d like to encourage a TPF port, which though interesting, exists in a parallel universe that is not a good match for the model. My experience with porting to Linux under z/OS informs me that there should be little, if any, problem with getting Ruby running under it. Access to more mainframe-specific technologies, such as Hypersocket SC/EE.

I was somewhat dismissive of the idea of the Ruby community speaking at “IT Executive” conferences. Ruby is entirely too early in the adoption cycle for that to be the appropriate thing. Java didn’t show up at that venue until well into it’s late-adopter cycle, which Ruby is in at this point. Familiarity with adoption-cycles really should make this obvious.

Finally, he wants answers to a few questions:

Does complexity increase within an enterprise that already has multiple languages and they attempt to bring in yet another language called Ruby?

Who knows. It depends on the capabilities of the organization to integrate ideas and models. There are already dozens of languages in the typical enterprise, and so the addition of another is of marginal distinction. Such a vague question is impossible to answer. Properly implemented, it may actually reduce the inherent complexity of solutions.

Have you ever wondered why James Robertson who has a vested interest in SmallTalk seems to always avoid talking about why SmallTalk is inferior to Ruby and instead prefers to amplify debates between me and the Ruby community?

Smalltalk (lower case ‘t’ please) is hardly inferior to Ruby. In many ways, Smalltalk, like Lisp, is superior in almost all technical aspects to current popular languages. Ruby took many of its ideas from Smalltalk, and acknowledges that. Such absurd statements are unfortunate, but demonstrate a lack of perspective. The “failure” of Smalltalk was primarily one of marketing, not technology. Would you argue that marketing is more important?

I would like for you to consider for a moment that not all EA’s and what they conclude at work necessarily believe their thinking when at home. In my travels, I have met lots of folks who are enterprisey according to their day jobs yet use Ruby at home (many are on my blogroll). I have also asked them that if we both collectively agree to the merits of Ruby within large enterprises, how come we don’t talk about it at work? What do you think the response to this question is?

Money. Preservation of power. Group-think. It’s human nature not to want to take a risk. It is exceptionally rare for anyone in the “enterprise” space to take a risk that might get them fired, even if the upside is potentially huge.

You are probably aware that I am one of the biggest on the record enterprisey contributors to open source, community-oriented and charity-oriented individuals you will ever meet in the blogosphere or even in person for that matter. Would love to know why you think I have taken such a position in talking about Ruby?

Do you believe it is out of spite, fear, evil, tough love, insight that others don’t share, etc? I hope you don’t say its traffic as I can statistically prove that blog entries commenting on the behavior of industry analysts and general topics on enterprise architecture drive way more traffic by factors than the mention of Ruby.

I have no idea what your motivation is, nor do I ascribe any to you. Motivation isn’t very interesting in the area we are discussing.

Conformity begets homogeny and is the antithesis of innovation. Never forget that.

¹ Why? ASN.1 is a painfully complex standard that has a lot of areas that aren’t well exercised. This means that the code often has latent bugs that can be exploited.n