Nobody likes to be called a dummy by a dummy.

Architecture Astronauts

I used to work with the HR-XML consortium about ten years ago.  After I left that job I lost touch with them.

This week I had a need for a job definition (we're doing some hiring) and went and looked at what the current XSD was like.

Man, what a incomprehensible mess!

They apparently got hooked up with the Open Applications Group, and now the XSD files nest 6 or more levels deep.  Which is an estimate, as Visual Studio chokes on them. You need XML-Spy to view them, which costs $500 for the edition that can create XML schemas.

Needless to say, I think I'll roll my own based on what I remember.
Permalink Send private email xampl 
April 6th, 2011 11:15am
Hehe I worked with the HR-XML specs for a while, and still get their various organizational emails. They have a remarkable number of conferences and get togethers for that.
Permalink df 
April 6th, 2011 11:46am
There's no money or prestige to be made with Keep It Simple Stupid.
Permalink Moneychanger Wannabee 
April 6th, 2011 11:50am
>> There's no money or prestige to be made with Keep It Simple Stupid.

Good God, frame this quote. A keeper for sure.
Permalink Bored Bystander 
April 6th, 2011 11:51am
We did this... ASCX12.... I swear I went to a meeting and the definition of "definition" became a topic for debate.

*head desk*
Permalink Jack's tautological turmoil 
April 6th, 2011 12:09pm
>now the XSD files nest 6 or more levels deep

Zoinks!
Permalink Some dude on the 'net 
April 6th, 2011 12:17pm
If you ever want to see data modelers gone wild, check out the SMEF framework from the BBC.

Unbelievable.
Permalink Moneychanger Wannabee 
April 6th, 2011 12:19pm
"There's no money or prestige to be made with Keep It Simple Stupid."

Early contender for quote of the year.
Permalink Lurker Indeed 
April 6th, 2011 12:42pm
The idiot peer who wants to pad his resume has been the death of several projects I have worked on.

Also, many simpleton owners and managers are impressed by layers of bullshit. They simply assume that a focused professional who does not blab about his "research" is doing nothing of use. This industry runs on hype and dishonesty.
Permalink Bored Bystander 
April 6th, 2011 12:45pm
That's because it can't run on reality. The reality of software development is that management CAN'T HANDLE THE TRUTH!

Heh... Seriously.

I understand they want to see quantified results. But the fact is, software isn't quantifiable in the way sales or support calls are.

Even assuming you use some kind of rational and sane approach to project management and estimation, you will be wrong, or you will be lucky. You might be close, but even if you luck out and hit it dead on, the merits of the product itself are subjective.
Permalink Jack's metric machinations 
April 6th, 2011 1:05pm
Fortunately, at least one guy with prestige sees the point:

http://tirania.org/blog/archive/2011/Mar-29.html
Permalink Lurker Indeed 
April 6th, 2011 1:21pm
Dependency Injection does seem to be the fad these days - a perfect example of a solution looking for a problem.
Permalink Lurker Indeed 
April 6th, 2011 1:22pm
>> The idiot peer who wants to pad his resume has been the death of several projects I have worked on. <<

I've seen Resume Engineering at a couple of places.  Usually it's one guy who is just drooling for the next release from MSDN so he can install it all. 

As a consequence, nothing he writes will ever build for anyone else, as the references are all borked up.
Permalink Send private email xampl 
April 6th, 2011 1:26pm
And the "latest and greatest" usually requires a lot of tweaking to get it to work.

I'll never forget the time cygwin 'promoted' their X11R6 to X11R7, and moved ALL the include files and ALL the lib files OUT of the X11 subdirectory, and into the global "include" and "lib" directories.

NOTHING built after that.
Permalink SaveTheHubble 
April 6th, 2011 1:28pm
Lurker - the simplicity thing is fine... the jump in feet first and start writing code on a $1M+ system with minimal architectural work is Fail.

Although, I am reading into what the guy actually said a little...
Permalink Jack's metric machinations 
April 6th, 2011 1:44pm
"That's because it can't run on reality. The reality of software development is that management CAN'T HANDLE THE TRUTH!"

So true. Management wants to know how long something will take to develop and how much it will cost. Not unreasonable to want that, but impossible to know with any accuracy in advance, but MW's Law #2: the more you know what you want to develop, the less you know how long it will take and if you know how long you have to develop, the less you know what you will get.
Permalink Moneychanger Wannabee 
April 6th, 2011 1:50pm
>> the jump in feet first and start writing code on a $1M+ system with minimal architectural work is Fail.

Agreed 100%. The specific mistake I have seen is for managers to confuse architectural work with pure bullshit.

An example of pure bullshit is the solution in search of an application. Trying to inject tools or methodologies midstream with no respect to the project goals.

I've even seen entire software designs based on astronautical architecture. The lead had a fetish about some buzzword so that buzzword pervades the design and distracts everyone from the project's real design or purpose.
Permalink Bored Bystander 
April 6th, 2011 1:56pm
With standards everyone wants to get their fingers in the pie so they can claim on their CV they were instrumental in putting some inane feature in the spec. This sort of crap destroys specs.

C++ committee finalized the '0B standard. Was 0x from 1999 to 2009, then they had to call it 0B since it took more than 10 years.

Now, do we have even ONE implementation of the PREVIOUS C++ standard that is complete? NO! Why? Because it is impossible to build a compiler to the spec!
Permalink Idiot 
April 6th, 2011 2:15pm
I think there is one :)
Permalink Vaxen 
April 6th, 2011 2:20pm
One.
Permalink Vaxen 
April 6th, 2011 2:21pm
No, really, it's impossible.

http://yosefk.com/c++fqa/

http://yosefk.com/c++fqa/defective.html
Permalink SaveTheHubble 
April 6th, 2011 2:28pm
What a great FAQ. No wonder I am so weird. I used C++ professionally for almost 20 years. That has to cause dain bramage.
Permalink Bored Bystander 
April 6th, 2011 3:12pm
Hm, OK, looks like they do support C++, it was C99 that is not complete.

http://gcc.gnu.org/gcc-4.5/c99status.html
Permalink Idiot 
April 6th, 2011 3:15pm
BB, the FQA is the greatest reference to C++ in existence and I consider on par with the K&R for depth of understanding of language issues.

When the FQA was discussed on JoS some years ago, the author responded to questions there and got tons of abuse from the ignorati. But the simple fact is that the FQA contains more understanding of C++ than any other reference and its author is the only person living who fully comprehends C++.
Permalink Idiot 
April 6th, 2011 3:16pm
I am thinking about this:

http://www.edg.com/index.php?location=faq_q1_whatsell

Unfortunately it's not really a compiler :)
Permalink Vaxen 
April 6th, 2011 3:20pm
"But the simple fact is that the FQA contains more understanding of C++ than any other reference and its author is the only person living who fully comprehends C++."

He is probably not the only person, but other than that you are 100% right.
Permalink Vaxen 
April 6th, 2011 3:21pm
>> the jump in feet first and start writing code on a $1M+ system with minimal architectural work is Fail.

Total agreement.  It's when you get into a solution looking for problem zone (i.e., what if we decide to move our data warehouse to a completely different platform when we currently have no reasons to do so?) that I have issues, and that seems to be where most astronaut types go.
Permalink Lurker Indeed 
April 6th, 2011 3:26pm
I'll mention a team that wrote its own death warrant by switching stacks mid-project because a lead contractor sweet-talked them into it. He bailed out, upwards - the rest were axed.
Permalink trollop 
April 6th, 2011 7:37pm
And some recruiter got a boner over his now-enhanced resume and he is off to a higher salary and bigger things yet, right?

Saw the movie, got the fucking 3-D glasses... :(
Permalink Bored Bystander 
April 6th, 2011 11:06pm
>>
what if we decide to move our data warehouse to a completely different platform when we currently have no reasons to do so?
<<

Favorite story from a previous job.  New CTO was hired, and was concerned that we used the MSFT toolset.  Called it a "monoculture".  So he convinced the board to spend $3 million and bought everything that BEA makes.  Plus a team of consultants to help us make the switch.

When they finished with their proof of concept, running it on a very powerful machine (for the time), it scaled to a grand total of 2 concurrent users.
Permalink Send private email xampl 
April 6th, 2011 11:08pm
So yes, I am now very gun-shy when it comes to someone making sweeping architecture reasons without first convincing me that it's truly needed.  And I'm a hard man to convince.
Permalink Send private email xampl 
April 6th, 2011 11:09pm

This topic is archived. No further replies will be accepted.

Other topics: April, 2011 Other topics: April, 2011 Recent topics Recent topics