Sharp as a Marble - Near death

<< Dyin' | Home | Almost there >>

Near death

It's brutal. This training is awe inspiring in it's complete cluster f*ckness. We only have these consultants until Friday at 2 and today they spent most of the time troubleshooting production code.

After 3 days, I finally got my system to hit a break point. It took 3 full days to set up a development box. They piled on STRUTS & SPRING and J2EE and every last framework they could find and unless you've been living and breathing this stuff for 2 years, it makes not one lick of sense.

The burning rage inside me is because I know I could rebuild it from scratch in hours and I know my job is to support, not to 'improve'.

We broke early today because it was a wash. With production issues, we couldn't go through the code at all. My brain hurts and we've still got 2 days left.

By the way, Ansible is almost ready for an alpha release. I've got a few bugs to shake out, then I'd like to put it out for you guys and gals to bang against to see where the weak spots are before I go completely live. Hopefully by the end of next week I will be switched over for good.

Thanks for hanging in there, y'all.

posted @ 2/25/2009 5:40:28 PM

Comments

Justin Buist says:

on 2/25/2009 9:26:34 PM

They piled on STRUTS & SPRING and J2EE and every last framework they could find and unless you've been living and breathing this stuff for 2 years, it makes not one lick of sense.

But it's so extensible! :)

BTDT... although I think my experience was a whole lot better than yours is going to be. I got on the project long enough before it went live to get a handle on everything, and it wasn't that big of a project when I came online -- only about half a million lines of code, most of that pretty generic boilet plate stuff.

Welcome to the world of Java. Why write application code when you can just drop another framework on top and push all the logic into an XML file?

Robb Allen says:

on 2/25/2009 9:34:06 PM

Welcome to the world of Java. Why write application code when you can just drop another framework on top and push all the logic into an XML file?

I am so sending that to all my teammates

Nomen Nescio says:

on 2/26/2009 3:33:19 PM

that description is genius, Justin.

all of this perfectly matches my own experience with J2EE as well; oodles of layers that don't need to be there, all the config pushed off in XML formats far better suited to be read by machines than written by humans, and everything seeming to spend all its time calling itself instead of doing the actual job. every time i start to think my own perl-and-javascript hacked-together website code stinks to high heavens, i take a look at JBoss and realize that no, i'm really not all that bad.

DJ says:

on 2/26/2009 8:33:39 PM

Robb, you might find this little story a mite funny. It sounds as if you need it.

I have a friend who spent a career as an aerodynamicist at McDonnell Aircraft in St. Louis. He simulated aircraft flight via Fortran software running on various mainframes. Remember now, this story concerns the mid 1970's to the mid 1990's.

These were the days of Analyzing Data By Looking At Plots, so he made it easy to make plots of the result of his simulations. The simulation outputs were files with headers followed by data, the headers describing the independent and dependent variables in the data. It was thus a simple matter to generate plots of data using a batch process; just edit a text file describing, for each plot, the file to extract the data from, the dependent variables to plot and the independent variable to plot them as a function of, the scale factors and labels to use, and so on. Cut it loose and out come the plots. In its day, this was as efficient as it gets.

One day, The Boss descended from his throne and stated that he had a presentation to give and his troops had one week to get his plots ready. There were 36 plots required. By god, they were gonna be done by hand, because he didn't like his troops to play with any goddamned computers. And the plots had better all look as if one person made all of them, so get your asses in gear!

The work was done by nine people, each making four plots. It took about one day for each one to make his four plots. It went something like this:

Round 1: "No, goddamnit! All of you use the same symbols for each data point of the same variable. Do 'em again!"

Round 2: "No, goddamnit! All of you make the labels the same! Do 'em again!"

Round 3: More of rounds 1 and 2.

Round 4: More of rounds 1, 2, and 3.

Round 5: "It's Friday. I leave in two hours. Are they ready?"

Well, they were ready, but he then discovered that he had told them to use the wrong data set, meaning files from the wrong simulations, so all the plots were of the wrong data. It was his fault, not theirs, and he was in a blind, raging panic.

My friend stepped up to the plate, saying something like this: "If you'll leave me alone for 30 minutes, I'll have all your plots ready with the correct data. Deal?"

The Boss nearly blew a gasket, but he agreed.

It took my friend about ten minutes to set up the plot run, five minutes to make the plots, ten minutes to duplicate them on transparencies (remember the year, this was done one at a time on a copier), and five minutes to gloat while the asshole picked his jaw up off the floor.

He never asked for hand-drawn plots again.

Comments have been closed on this topic.

Comments

Archives

Categories

Comments

Archives

Categories

Second Amendment

Sharpest Marbles

Blogroll

Funny, as in Ha-ha

Good Reads

Extended Blogroll

Industry Links