10 Reasons Why Being a Lazy Dude is Actually a Good Thing

Started by DigitalBuddha, February 03, 2013, 08:23:09 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

BikerDude

Quote from: RighteousDude on February 06, 2013, 12:36:39 AM
Quote from: BikerDude on February 05, 2013, 08:13:43 AM
I try to keep logic out of the DBMS when possible. Decoupling app logic adds complexity.
But stored procs often have a big performance advantage.

That's where I see the glitch. If I can poke a stored procedure into the database I can gain both a runtime advantage and a code (non-) complexity advantage. If portability is a concern I can use a fucking phrasebook that says use this query for this database, that query  for that database, and some really obscure shit for some really obscure database. In, out, done. Cheap, easy, good, pick all three.

Not so very long ago (from the perspective of a 50+ Dude) I wrote a middleware-ish application that tied big-name routers, an ubiquitous network security scanner, and an ERM system together. The interesting part was interfacing to the ERM, which has an API and whatnot but that requires developer seat licenses and all kinds of other bullshit that I wasn't enabled to wade through to get at it. We had us a big ole geeky motherfuckin' teleconference with some managerdoids involved to make sure that no fiefdoms were encroached, and the final solution was that the DBA's who understood the ERM would create some procedures, and all my code would have to do was spew some data at the database. No API bullshit, no seat licensing, no pain at all. Some DBA spent four minutes crafting a nice procedure, and I passed it some data. It was easy as jerking off manually with a palm that profusely sweats oil when aroused.

So my end just spews data, their end just twiddles it with DBA-fu, and that is that. It works so fucking well that they haven't had a beef in six years.

That's my perspective as a developer. If some DBA can improve my queries, fuck it, let 'em at it. I don't want to learn which SQL works with which database and which are more efficient. Let the guys who live and breathe that shit file bug reports. Meantime, I'll think about my code. Win  : fucking win. Put the logic into the fucking database where the guys who live and breathe that shit can deal with it, and in the end everyone wins.

For me the big thing is to keep DBA's out of the equation because in most large agencies the "process" suxs.
You throw a request over the wall and twiddle thumbs till it's implemented.
The goal for me is to implement the system so I can sit someone down and pull the whole project for Subversion or CVS and twiddle a couple configs and have a working system. Ownership is a huge advantage.
As soon as you start to spread that out you get anguish from other people's nonsense.


Out here we are all his children