I programmed a poker bot for one of my university projects. It could hook onto an online poker client, read the state of the game through a combination of pixel matching and OCR, and then used ANNs and targeted simulations to decide on the best action to take, and then click the appropriate button on screen. It took me about 3 months of programming to get it done.

Once I finished university, I decided to create a newer better poker bot. It was going to be amazing. This version would have a database to store previous hand histories, which could be used to create an individual ANN for each opponent it met, improving the accuracy of the opponent modelling. It would have a bot simulator, where bots could play each other at 500 hands a second, producing a nice graph and all the statistics you could need to see where their weaknesses / strengths were.

I spent the next 3 months designing it all. Classes, databases, all perfectly designed adhering to OO principles.

Then I started coding. Despite all my planning, problems popped up that I hadn't foreseen. No problem though, I'll just do a bit of a redesign. Designs were changed, classes removed, added, amended. Carry on coding. More problems. More redesigns.

In the end I spent more time designing than I did coding. It has been more than a year now, and my poker bot has less than half the functionality that my 3 month uni project had. I don't spend a huge amount of time programming - I have other responsibilities nowadays that I didn't have when I was a student - job, baby etc, but I think the real problem isn't lack of time, it is lack of motivation. I have spent so much time on the project, but have barely anything to show for it.

My advise - don't worry about design very much at all. Have a quick think about how the main parts will work together, and start coding. Dont worry about needing to rewrite code when things go wrong. The most important thing is to get coding and get something working to keep you motivated. Better a working program with terrible source code, than a great design and no program.

I programmed a poker bot for one of my university projects. It could hook onto an online poker client, read the state of the game through a combination of pixel matching and OCR, and then used ANNs and targeted simulations to decide on the best action to take, and then click the appropriate button on screen. It took me about 3 months of programming to get it done.

Once I finished university, I decided to create a newer better poker bot. It was going to be amazing. This version would have a database to store previous hand histories, which could be used to create an individual ANN for each opponent it met, improving the accuracy of the opponent modelling. It would have a bot simulator, where bots could play each other at 500 hands a second, producing a nice graph and all the statistics you could need to see where their weaknesses / strengths were.

I spent the next 3 months designing it all. Classes, databases, all perfectly designed adhering to OO principles.

Then I started coding. Despite all my planning, problems popped up that I hadn't foreseen. No problem though, I'll just do a bit of a redesign. Designs were changed, classes removed, added, amended. Carry on coding. More problems. More redesigns.

In the end I spent more time designing than I did coding. It has been more than a year now, and my poker bot has less than half the functionality that my 3 month uni project had. I don't spend a huge amount of time programming - I have other responsibilities nowadays that I didn't have when I was a student - job, baby etc, but I think the real problem isn't lack of time, it is lack of motivation. I have spent so much time on the project, but have barely anything to show for it.

My advise - don't worry about design very much at all. Have a quick think about how the main parts will work together, and start coding. Dont worry about needing to rewrite code when things go wrong. The most important thing is to get coding and get something working to keep you motivated. Better a working program with terrible source code, than a great design and no program.

I perfectly understand this situation. It happens all the time when you try to approach a task with too much analysis. Paralysis by analysis, http://en.wikipedia....lysis_paralysis

I know the OOP is particularly vulnerable to this issue. I say this because I have a feeling that OOP was overemphasized in CS courses. So much so that I find that there are so many problems that DON'T fit into an object-oriented based model naturally and we programmers spend so much time trying to "fit" the problem into the solution we have already planned out. I think the approach at schools and colleges is too theoritical and technical.

The key, I think, is to realize that OO is just another model of problem solving in general and is not a magic bullet for every kind of programming task. Sometimes I think C++ is the biggest culprit in bringing in too much complexity to the OOP model and people spend too much time beautifying and dressing up their classes rather than make things work.

Also the "verb" problem always exists in OOP: should it be thisobject.performTaskOn (thatObject) or thatobject.performTaskOn (thisObject)?

I work by making lists.

Make a list of things to do, with some criterion for knowing when one task is finished. Try to keep this list in logical order that things should be done, but of coursemeplay, you may reorder items later.

Example:

1. Get gfx initialisation, engine inintalisation code working
1. Write file format convert & write map loader
1. Write background renderer
1. Get object manager / engine working "enough"
1. Implement basic physics / gameplay
1. Implement object loader (from map file)
1. Implement some Things in the map (e.g. powerups)
1. Implement bad guys
1. Fix end-of-level, next level, player death etc
1. Tart up graphics
1. Create the shop for buying new weapons etc for next level
1. Add sounds etc
1. Further tart up graphics...

Then potentially divide tasks into subtasks. If you find that you're running out of time you can cancel tasks, or move them on to a lower priority list

Anyway, this is what I do every time I enter a LD48 or similar.