Discussion on:

Five common programming mistakes

SQL Server Management Studio's code completion can really get in the way. Type a short keyword, then space, and you can get something totally inappropriate. And, I always seem to forget to press escape to tell it to let me have what I want.

Same issue here. Some days it gets to the point where I just turn it off.

It does require some interesting keyboard gymnastics. I find myself writing the FROM clause first, then up-arrow-ing back to the SELECT clause... a very un-natural way to write code.

For every stupid mistake it keeps me from making, it seems to introduce two more.

Trying to make #1 clear to my VB class, but they just don't get it. They are of course learning constants today, so more issues will come about. Yay...

When I used to teach (in those days Windows (1) ran from a 5 inch floppy and the 80286 processor was golng to solve all computing problems because it could addess a whole megabyte of RAM) I would have the students create a fairly complex piece, then work on something different for a while, then come back and make some changes to their original code. That got the naming and style consistency points across.

The delusion that the code will not change.

Everything else involved in the creation of any piece of software follows from that.

I've talked about this before on here, but it may be worth repeating. There's a temptation to lean on an IDE's application support features to design your app. for you. I discovered years ago this was a big mistake. Beyond editing, compiling, and debugging, use the IDE's features with caution. You could paint yourself into a corner.

what dimwit thinks a DESKTOP application does NOT need any menus at all?
[ Chrome being the perfect example ]
what shyte for brains thinks that 12 instances of a graphic widget uses less resources than 1 graphic widget instance with an array of 12 pieces of text?
[ Chrome's address bar inside each tab being a perfect example of that idiocy. ]

a web browser that makes it extremely difficult for you to open a file on your own computer to view it by not having a menu to give you access to that functionality is 100% useless.

any application that changes ui design completely, said changes increasing resource consumption, needs to be redesigned.

edited to add:
why is it that our thousands of times faster computers, with millions times the resources, run so much slower that the older systems did?
I recently got an i386 notebook, 16 mb ram, no cdrom, no space for a battery at all, black and white 800x600 display.
the freaking thing boots, loads ANY application in far less time than a dual core system with 2 gigs of ram.

Whenever we do programming someone stuck at the somewhere and start conversation without checking the logs.... it's not the character of any programmer before asking for help - we have to try all possible ways to recover from it.

After the last compile.

Plus overdependance on external libraries

Hi,
once I had to adjust timestamps in a log analysis. The adjustment was variable in duration and direction. Could be 'add 35 minutes' could also be 'subtract 56 minutes'. None of these operation is simple as you can change day, and/or month and/or year, not excluding leap years. The clue has been to convert original tiemstamp into a number of days and hours, apply the change and convert back to mm/dd/yy(yy) hh:mn:ss. For that kind of convertion the DATE routine of REXX has been a great help.
Regards, Jacques

One of the biggest problems I see in some newbie programmers is wanting to start coding before they even know the whole problem. You have to spend judicious time in the design phase BEFORE you start coding. (OK agile enthusiasts, you can start hammering now...:-) . There are very few occasions where I can look back on a project and say "I wish I would have spent less time on design."

I get asked to make apps and no-one has really thought of what they need, just what they want (not the same thing) so prototyping is the only answer. Yes not the same issue you are having but a lot of people seem to think that I am doing what you describe. Just remember, never use any of the prototype code in the final, it will always bite you if you do.

Last time it was just to design a jeopardy like game.
Luckily they accepted my first design, but I went overboard in making it customizable, every image, shape, text, everything was controlled by INI files and the questions/answers and scores, players, etc. all stored in XML.

It will be version 1 before you can realise what a terrible error you just made.

You can design by having a go.
The real problem is when you use this approach you need to be ruthless. If it turns out your design is lacking, bin it, and use the lesson. Don't rescue it, don't patch it, don't leave it for version 2.

Now in the commercial software world, when v1 with it's technical debt has been delivered, what has actually happened.

While I wholeheartedly agree with you that time spent designing before you put your two fingers on the keyboard is time well spent, businesses in the main do not share our opinion. As long as they can sell it, it will do. Spiraling maintenance costs and no longevity are out of scope.

It's another area where academia could improve as well with some more real life scenarios, than swap this bubble sort for something more efficient.

I can't find it on Google or Bing. Has anyone got a link to this?

Some programmers using Oracle's PL/SQL languague use explicit strings for dates and when execute the code the dates are wrong because they forget that explicit date convertion are different on IDE (TOAD, DreamCoder, SQL*Plus, SQLTools, etc) that on the server where the Oracle instance are running.
I always recommend to use the function to convert string to dates (TO_DATE) and use the second parameter to format the string:
TO_DATE('2012-09-01','YYYY-MM-DD').

I wish we could just check the log, but it just outputs the same error you got on screen. Useless.

If the log file can't tell you what is happening at the moment before the error message, you might as well not bother.

I remember dealing with this design issue when I first started into client/server software. Somehow I got acquainted with the kind of people who were using the client software (every application developer should do this), and I realized that the detailed error messages I was putting in the log were not appropriate for them to see. I wondered, though, "What would be appropriate?" I worked this through with my fellow developers on the project, and we came up with about three error messages that were "good for all occasions." We'd send those back to the user, using a little discrimination about which one to pick, and we kept the detailed error messages on the server log, for us, or the customer's administrator to see. If it was something that the user could correct, they were notified of it, and allowed to correct it. If it was a system problem, all they'd see was something like, "Transaction failed. Please contact administrator." What we realized was that the issue of error messages was really about delegation of responsibility, and then considering how much responsibility each person who might see it was capable or authorized to take on.

I would have hated to work on your helpdesk.

Yes, now that you point that out that could be awkward.

Possibly the wording of the error message could be improved without reducing available detail. Common practice is to put a prominent large font user-oriented warning but include technical detail in low contrast so that it's available if helpful.

Some users are more savvy and can possibly work out a course of action based on which area the fault occurred in. An on-site technician who doesn't have immediate access to the log may be able to solve immediately if the fault is clear. Also cross-referencing the particular error for any one user to a large volume log may be non-trivial.

I wonder if this post indicates a 6th type of programmer error: not consulting users on features and behaviour? Mark says "know thy user" but he then says "I realized" and "I worked this through with my fellow developers". No other stakeholder consultation involved?

Sometimes errors are not really errors, like an expired session that is reported as a database error, or a failure to delete is legitimate because there's additional referenced content, but this is not explained. But that's another problem and definitely requires direct communication to the user, not obscuring it.

Great work Justin, you have listed quite a number of what we face almost everyday-I guess some tools help out more than others in different aspects you have mentioned. However, I think this list is too short So I am eagerly looking forward to a part 2. Probably titled: 10 very common mistakes made during programming!

This isn't even programming 101. Programmers need to accept that diligent detail to the level of paranoia is required all day every day every year they'll be programming. If they can't even apply consistency to variable naming they haven't got the right attitude. We all make mistakes but you'll make a lot more if your approach is wrong. If these mistakes are common we have a shortage of good programmers.

That there's a demand for good programmers....

I am constantly amazed at how many think this a good idea.

It is lazy programming, and evidence of complete lack of planning.

Line 1 error, so lines 2-20 are useless, skip them and go to the catch statement.
Or function error, exit with a failed return code.

We all know that exceptions are supposed to be used for exceptional error conditions, but I think he was saying that rather than, say, returning from a function once a process is done (no errors), the programmer throws an exception instead to alter the flow control in the function that called it. So rather than using a conditional in the calling function to test a returned value, the programmer sets up a try-catch in the calling function that basically works like:

try
{
askUserForSelection();
// If they pick option A, execution will begin here
}
catch(choiceException)
{
// if they pick option B, askUserForSelection() will throw this exception,
// and execution will begin here.
}

Another "idea" is to use an exception to break out of a loop, even though it's a lot easier to just use a break command inside the loop.

Documenting the obvious (e.g. increment counter) whilst not documenting what is the purpose of a variable or segment of code. Most of us can determine WHAT the code is doing but not perhaps WHY it is being done.

Proving you are clever by writing a C program that generates the 12 days of Christmas.

Nesting because you can, not because it is necessary.

Remember folks, someone else someday will need to maintain your code and you really want that person to say "What a beautifully written program" not "He/she was either genius or an idiot and either way I hate this program."

Yes, sometimes web developers are just too excited about the UI components capabilities that they use it inappropriately to make the website sophisticated and hopefully give that 'wow' impression to the visitors. But instead, it creates confusion and visitors sometimes find it not user-friendly. My advice is to make it as simple and straight forward as possible, and use those special UI only when it adds convenience to the users.

And as for variables naming, have your own standards and make sure you and your teammates follows them strictly. If you use "cust_id" for customer identification number, use "cust_id" throughout your coding, may it be in javascript or php as long as it refers to the same item.

Hi, in case you have an error that you are trying to find in your code, here is an article I wrote as a general guideline for finding it:
http://efficient-work.blogspot.ro/2012/08/find-error.html
Feedback is welcomed!

I'd have worded it slightly differently, but the concept of not changing things until you know what you need to change, is a concept many should take home.

Ever worked with someone who corrects bugs by successively fixing symptoms?

for(int x = 0; x less than max; ++x)
...

In regards to the 1st point of the article, is 'x' correct being x? (or any letter of the alphabet for that matter...), because I've noticed that this is always the case... and y 'x'? who started this???