Coding Horror

pauldiv wrote: »

lol I bet it made you a much better sys admin though

Yes, but somehow I think the clients would have preferred I not be given that lesson at their expense!

Their website's performance.

What is up with SQL?

Almost every time I encounter SQL written by other developers, it seems that it has been somehow considered exempt from all rules regarding coding style.

Indentation and structure? apparently "its all one query" means "no indentation required"
Casing? SQL is case-sensitive, right, so there's no need to worry about case. There's no need to even worry about consistency of case.
Comments? Are you mad? That nested set of n functions doesn't need a comment.
Aliasing? Aliasing table names to a, b, c and d is plenty, right?

What's amazing is that most of the people who I see writing sql write really nicely structured code in Java or whatever....but as soon as its time to hit the DB, then its Sin City.

Sparks wrote: »

SQL also seems to be immune from even efforts at version control and deployment controls. The number of times I've seen a dev team working on one SQL codebase only to find there's a different codebase on the live server...

Oh, don't get me started.

Not really a coding horror, but once saw a programmer turn their app they were working into one line of text to protest someone taking over from them. Took them a couple of hours. They also gloated about it when the new guy turned up.

They died a little inside when the other guy pressed CTRL-SHIFT-F (auto formats the text).

Well since we're on the database thing, we've 3 core databases that support the concept of a financial agreement (FYI: a proposal is an unactivated agreement).

On the oldest 20yo system, the primary key on the mast table for instance is agno (most of the tables are limited to 4 character names, so table names like mast, mas2, cust, payh, dats, deco, ucod, sprm, and my personal favourate - tits).

Under the proposal capture system, you will generally find proposal_no, propno or agno. Under the MIS system you will find agreement_number, and under the collections sytem you will either find agreement_number (because of the feeds from MIS) or agno.

select blah from ifs_agreements a
left outer join agreement_status s on (s.agno=a.agreement_number)
left outer join proposal p on (p.proposal_no = a.agreement_number)
etc.

D.

A place I worked before had this developer - he was rubbish - one of those guys that got into software when companies were hiring anyone who could spell computer.

Anyway this guys is working on a large web app, web pages,com objects,database,etc. He just manages to finish the project (after about 6 month work) before leaving the position. When we go to actually deploy the project we find that it will only run on his machine! Somehow this guy has managed to hardcode his own machines details into the web app so it wont run on any other machine.

There was this other guy who joined the company and was given the task of writing a bunch of stored procedures for an application - just general stuff and reports. All in all there were about 60 stored procedures. The guy is found out to be a dud and let go. (was hired cus he was a mate of one of the managers). Anyway when we finally get around to checking on the work he has done, there is no SQL - its all just comments. So we have 60 SQL procedures just filled with comments on how he was going to select this from that, but no actual SQL.

Didn't know where to post this.

Reusing code time and again to workaround the regional setting and non Daylight Saving setting on the internal server in the room next door.

BostonB wrote: »

No meaningful information in the error handlers, so that users think the system has a bug/crashed, and they log a support call each time.

Aha yes, catching errors with an empty handler - even deleting the autogenerated "e.printStackTrace()". Error handlers that don't handle the error, so things appear to have worked and didn't work, and you have no idea why.

COYW wrote: »

These tables were created by both internal staff and external "consultants".

Fixed that for you.

I work in a sh1thole (commonly referred to as a financial institution). So we get this guy in that was just a horror, and after a few months I beg for the guy to be fired, which of course doesn't happen, and lo and behold we're left with 2 years of mess.

Coding horrors don't just happen in isolation, you need someone with an almost narcissistic belief in their skills and a matching aggressive streak just to liven up things. Like when I needed to run thru' code like this:

if (condition == true) 
 some block of code
else
if (condition == false)
 some block of code
else
 the block of code I needed to run

... it takes me 30 seconds to see to the problem (it was about 20 lines of code), and involves a 30 minute argument with the owner (who always kept their code - and half the rest checkout out to themselves) - eventually shouting at me "WHAT DO YOU WANT ME TO DO" even though I've explained it 5 times at this stage. Never trust anyone who can't read their own code

When the guy had been let go (I wish they had let me go and kept him - they're a much better match), I found myself cleaning and commenting a lot of the system - those bits I could. All these comments would have been on his code.

  //[dd] [functionxxx] is supposed to be a class function, but it is only called 
  //     in an instantiated version of the already existing class, and it returns 
  //     a second instance - real stupidity - its leaking like a rusty bucket. 


  //[dd] 30.10.2007 This is so broken its not even funny 
  //                I've depreciated due to the reference to PropList 


  //[dd] For the time being - just find the respective node, and pretend it has 
  //     been clicked. Its wasteful but... what the fcuk are you to do in this 
  //     mess? really??? 

  //[dd] 16.11.2007 Removed Delete Proposal - WTF was this added in the first place? 

  //[dd] 25.09.2007 [function name] - WTF is this in [unit] and not here??? 
  //                It doesn't appear to be referenced anywhere else, I've moved it here 

  aNewPanel := CustomerPanel; 
  Ntype := 'CUST'; // <--- wtf is this about? ntype? 

   {----- comment--- 
     This method below was required because of problems with *** 
     sufacing message record changed by another user etc 
     This method is used after a save to resolve the *** issues. 
     ----- new comment 
     there probably isn't any problem - its just the screen has 
     been written by an idiot. 

  //[dd] FFS popping up a messagebox inside a transaction is fupping criminal 

  //[dd] 19.12.2007 THIS SHOULDN'T BE HERE FFS. 

  //FFS main. <- this is sick 

This one is interesting, because as the framework was coming together there were stubs written that had no proper implementation at that time. One was focused on in a meeting and Narcissist coder freak comes out with the classic line "Oh you'd do that in the base class". I cringed, I protested, but they had their easy answer and he did it - and what a mess. This is a comment around code that didn't call the implementation in the base class.

  // sink can edit because it doesn't work for us here - fckn idiot 
  //Inherited CanEdit(bEnabled); 

D.

Hellm0 wrote: »

I think my least favorite thing to hear in a job is the word "Agile". The moment I hear the words "We are an agile shop!" I know to expect nil documentation, no concept of architecture and unreasonable deadlines.

You should apply for this job:
http://www.irishjobs.ie/Jobs/Visual-C-C-Engineer-Dublin-6299664.aspx

zynaps wrote: »

Oh man, "agile" and "Visual C/C++" together in a job ad do not promote a sense of wellbeing.

Also, "Min 2 years recent experience in Net Technologies". What's that? .NET? Networking? Maintenance of fishing nets?

Also also, where do they get this "min 2 years recent experience of <random half-desired skill>" from? I wish people would stop saying things like that in job ads :rolleyes: Back when .NET was about 4 years old, I remember seeing an ad demanding "min 6 years experience of .NET technologies" or such. :mad:

Yep. But it's hardly new; happened with Java as well.

One of my pet peevs:

// We should never get here.

Codpeas wrote: »

One of my pet peevs:

// We should never get here.

That's not quite so horrible as you'd imagine though, it can be good practice. For example, from a switch statement in a freeRADIUS cllient:

	default:
		/* Should never enter here */
		RDEBUG("Internal sanity check failed on eap_type");
		rcode = 0;
		break;
	}

lil_lisa wrote: »

I guess I should be grateful that my boss lives by the "Copy & Paste" rule...except for the fact that its his answer to everthing!

constantly having to perform a diff on methods that are usually about 100 lines longer than required and vary by 1 line.

new feature.. f**k it we will c&p feature x..

No code but software FileNet

Can't handle concurrency. They put a wrapper around a COM Object and said it was a .NET version. "Best Practices" is a laugh. Worst of all is they charge a fortune