What do you do when you're stuck on a problem?

[Aspirer]

That's the question: what do you do when you're stuck on a problem in programing? Especially when you're one man trying to learn the craft? I don't like asking for help unless its about something I'm sure I don't know anything about because that isn't learning. That's having someone give you the answer. I've been stuck on a problem for three weeks, with a little progress. Just got what I wanted my program to do working only to have another error crop up somewhere else.

I just take a break usually but that doesn't ever seem to help. I think that if I take a break at the appearance of every error I run across I'll never get anything done. In fact that seems counterproductive as programming is all about finding solutions to problems.

Oh yeah and you don't know anyone who knows anything about programming. I live in a town with no high school programming courses or college courses. I moved here from out of state. FML this sucks!

[DanielSefton]

I fix nasty problems through sheer determination and trying literally everything. I'm not the type of programmer that moves onto other things if something is taking me too long, I attack it until I find it. The problem with that is it increases project time.

You may not think that breaks help, but they do. The back of your mind may even solve the problem for you while you're away without you knowing, and when you re-introduce it to the code base you'll suddenly realise "OHH THAT'S WHYY". You might also want to shower more often.

Usually it helps to narrow down the problem by removing code until it no longer occurs. If it's a crash (ones without any useful stack trace are the worst), it's important to know exactly where it's occurring. If all else fails I use print statements everywhere (also because debugging in Xcode sucks). Scope problems are always a killer as well, be sure that things are staying in memory.

I was also in the situation where I had no help; in my town nobody understood what I was doing. The benefit is that you learn how to solve problems yourself.

[bstone]

Paraphrasing the old saying "The bug is where you find it". Beat it to death. It won't fix itself. Try to google for similar issues but with the goal of learning how people track them down and fix'em. Figure out the tools that are good for the job and learn to use them well.

[Zonder]

As DanielSefton said your mind usually hasn't given up and will be working on the problem. If you haven't encountered this yet you probably will at some point. But breaks are really important go do something mundane like make a drink mow the lawn etc. I have programmed since I was 8 after a while I realized I didn't think on stuff much anymore and do it automatically it's just getting to that stage.

Now one thing not mentioned is you may be lacking some information and that's why you can't solve the issue and you do need some outside input to understand. Asking for help isn't a bad thing to do sometimes as long as you make sure you understand why it went wrong. This is also what back to basics forum is for

[Klaim]

Wow I was thinking I was bad because I spend a lot of time solving hard problems and learning a lot about them but apparently I'm not! As DanielSefton said it makes project dev longer but I think persistence in willing to understand fully a problem (which mean you reach a point you can solve it or avoid it) is really important.
Unfortunately I tend to choose really hard problems so my projects suffer from that but I'm always motivated so it's ok.

Breaks indeed, also it have been discovered by scientific researches that the part of the mind that is capable of solving really complex problems is not the concious part. It is better to try to understand as much as you can about a problem, then stop everything and go do something totally different, like taking vacations. Your mind will indeed find more things while you're not actively trying to search for it. This explains basically why most creative people are creative because they learn how to take breaks often or maybe more like how to do things that have nothing to do with a hard problem so that it becomes less hard when you get back to it. It's a bit like cooking epiphanies.

You should search about creativity (which is the core of problem solving) if you want to know more.

Also, if time is counted, ask experts. I recently asked advice from Sinbad about a hard problem (something related to API in highly concurrent cases) because I knew he had faced a similar case with SourceTree and I was stuck for several weeks (also part of it was analysis paralisys). Sometime the problem is solved but you just don't know that it is, that's when pointers form experts can help. Maybe don't ask for solving the problem exactly but the general problem around it.

[TheSHEEEP]

This is what I do, usually:

1. Look if someone on the internet has had the problem before.
2. Ask someone who might know more about the problem than you do.

If none of this yields any useful results, you usually have to decide, taking into account how urgent something is or how important in the big picture):
Either, you implement a workaround, which may be ugly, but get the job done.
Or, you dig into the problem, which has the potential of wasting a lot of time.

To me, results matter far more than the way they are achieved, so I usually go with workarounds.
Only if that is not possible or if I really have enought ime at hand, I dig deeper.

I know a lot of colleagues who cannot resist the urge to understand everything and have to tackle anything from a pretty scientic perspective.
And that IMO is simply a waste of company resources. I know, I know, this sounds pretty unenthusiastic, but after all, when working on an actual project you are paid for, time is money and solving a problem should have a higher priority than understanding in-depth what's going wrong.

I don't know how often I saw how a colleague tried to dig deep into a problem, jumping from source file to source file, trying to grasp the big picture. And then I ask "Have you googled it?" "Uhm, no, I try to understand what's going on." Then I google the problem, usually find a solution or at least a very good hint and the problem gets solved much faster.
IMO, too many programmers have that strange desire to solve every puzzle themselves with as little help as possible.
And I really don't like puzzles

Of course, for personal projects and jobs that you do not do for the money, things are different.
Also, in some cases, it is obvious that a workaround might cause more trouble in the long run.

[Klaim]

Yeah two fun things about what you said:

1. as now I'm not in a company (I'm working for myself but not making money yet) then I was thinking about personal projects only where I tend to prefer to understand why something don't work simply because most of the time I don't have a deadline - but really I should! Anyway I still search online and ask on stackoverflow first if the problem is really about knowledge and not a more domain-specific problem;

2. When I was working in other companies I had the same experiences of helping coworkers solving problems in less than a minute with a simple google search. It's incredible how many devs don't have this reflex of searching online first when something strange happens but I guess it comes only when you realize you are stuck. But in my experience, even if you are not stuck but are hitting a minor problem that you don't see immediate solution, it's faster to see what's the knowledge about it online first, then decide what to do. That being said I often code without connection (when in the train for example) and in these cases I still need to be very persistent on hard problems (or wait until I have online access again);

[Zonder]

You will never work on projects which have a locked down network though IE no internet access at all. You would have to wait till you left the building to search for it to solve the problem. So knowing how to solve problems is very important.

[TheSHEEEP]

You will never work on projects which have a locked down network though IE no internet access at all.

Wait... what?
Why should one work in such an environment?
Where does such an environment even exist?

Of course you are right that knowing how to solve problems is essential, but... that got me confused

[Zonder]

Why should one work in such an environment?

Well the why is don't if you don't like the restriction

Some areas where it can happen are in Government Agency / Military / Research. There the only ones I can thing of but it does happen.

[bstone]

Even putting the most restrictive environments aside, companies will likely hire problem solvers instead of agile googlers.

[TheSHEEEP]

Well, that implies that you see a difference there.
I would say agile googlers tend to be the better problem solvers, independent of their googling success.
Using available resources (like the internet) is just one of the things you have to be good at.
I don't know too many people who are good at "online research" to solve a problem, but would not be able to dig into the problem themselves.

[bstone]

Well, look at the facts. Here on Ogre's forums the googlers struggle a lot with relatively simple stuff, getting stuck whenever they tap an area without a ready-made solution. And I believe a googler's mindset is a bit different from that of a problem solver's. Problem solver will be able to google stuff up, not an issue. But that doesn't work the opposite way from what I can tell.

[holocronweaver]

When I am stuck on a problem, and additional work offers no revelations, I usually switch to a different task for the rest of the day and get a good nights sleep. Often I wake the following morning with a solution or at least an idea.

I agree with DanielSefton that showering can really help by releasing dopamine and serotonin. The same neurotransmitters are released during exercise, but they have a smoother increase and last longer throughout the day. I cannot problem solve nearly as efficiently when I skip exercising for too long. Even taking a ten to fifteen minute walk can get the blood rushing, elevate your mood, and make you see things differently. Eating healthy is also important. When I regularly skip meals or eat lower quality food, my output gradually degrades.

Well, look at the facts. Here on Ogre's forums the googlers struggle a lot with relatively simple stuff, getting stuck whenever they tap an area without a ready-made solution. And I believe a googler's mindset is a bit different from that of a problem solver's. Problem solver will be able to google stuff up, not an issue. But that doesn't work the opposite way from what I can tell.

A programmer who never uses a search engine to find solutions to common problems will waste precious time reinventing the wheel or leafing through textbooks. On the other hand, a programmer who never learns the basics and attempts to use search engines to skip ahead will eventually hit a brick wall. To me this is a problem that solves itself. Any committed programmer will, through trial and error, find a healthy balance of in-depth knowledge and reusing the results of others.

[TheSHEEEP]

A programmer who never uses a search engine to find solutions to common problems will waste precious time reinventing the wheel or leafing through textbooks. On the other hand, a programmer who never learns the basics and attempts to use search engines to skip ahead will eventually hit a brick wall. To me this is a problem that solves itself. Any committed programmer will, through trial and error, find a healthy balance of in-depth knowledge and reusing the results of others.

Yup!

[Kojack]

You will never work on projects which have a locked down network though IE no internet access at all.

Wait... what?
Why should one work in such an environment?
Where does such an environment even exist?

Of course you are right that knowing how to solve problems is essential, but... that got me confused

Any company that doesn't want to have a source leak (like Valve had with HalfLife 2) may limit internet access. While I've never worked at such a place, I've heard of companies where there were two physical networks, one for the local system (used for development, source repo, etc) and another for internet. If you wanted access to both, you needed two pcs. That way no virus or bad user practice would expose internal resources.

I know of a college (not mine, a competitor I won't name) that used to white list web access on all student computers (this was many years ago, probably different now). You could only visit sites that were pre-approved by the it staff. Only one pc per classroom had unrestricted internet, the teacher's pc. I guess it stops students using too much bandwidth with youtube and pandora, but would suck if you were trying to get answers on obscure coding forums.

Of course many people have smart phones now. If I had no internet access on a pc I'd just google on my phone. Or tether my pc to the phone, but if there's no internet on the pc it's probably due to a policy and I won't want to break that.

There's so many hacking stories popping up on places like slashdot that leave me thinking "Why the hell was that even on the internet!"

[Klaim]

Yup++

The googling is mostly to fix problems that don't need specific understanding of the situation, like when the code say something but does something else and in the end it's a compiler bug (that happen if you're "on the edge"...).
In my last job in an embedded software company I was promotted expert bug hunter because I enhjoy hunting very hard bugs into the guts of big projects (like the time I hunted a memory leak in Ogre just for fun - I really like to hunt memory leaks but I don't have much). They had limited internet access so I still feel the pain of having to use some tricks to be able to go on some very obscure forums to get the right information about the obscure libraries used.

You get the balance with experience, which gives you quick intuition on when to take the time to google. Also it's actually not my first reflex personally, my first one is to check if I understand my problem correctly or detect if I'm confused (which is not always instant).

I think the knowledge to understand how to use Ogre quickly is MASSIVE (almost modern C++ + how graphic cards works + specific graphic API quircks + what features are available in what version of Ogre) so I'm not surprised that a lot of people rely on googling. Also a forum isn't perfect for finding exactly the answer to a question specific to Ogre (which is why they did StackOverflow).

[mkultra333]

I think I can break problems into 3 classes.

1. How do I do something? (How do I get convex hulls working in bullet? How do I set up MRT without using compositors?)
2. Why isn't this working? (How come these quaternions are coming out wrong? Why is bullet crashing when there's lots of hulls? Why am I getting input lag?)
3. What should I do? (Should I build detail into the map or add it as meshes? Can I make robots out of lots of different meshes and put them together as the game plays?)

Type 1 problems are usually solved fastest with googling and forum questions.
Type 2 problems I usually solve with the debugger, Programming By Permutation, and getting the program to more and more verbosely log everything it is doing.
Type 3 problems I let float around in my mind, exploring different possibilities. I think about them when I'm not doing anything or going for a walk. I scribble down a few notes, look at a few numbers, sleep on it a few days. Usually a solution appears or the balance of pros and cons falls one way or the other.

[madmarx]

When I am stuck on a programming problem, if the debug session wasn't successful, then I can split, compare or use SVN. Also I develop my programs so that they can be splitted easily.
1/ I begin with a "functionnal" environment and I add things progressively (libraries, codes & so on), and try to get the bug back.
2/ I make sure if it's a game that I don't have to play the game to get the bug (eg : create a save state just before the bug). I create interesting break points. I use simplified data.
3/ I compare with previous SVN version when there was no such problem.

Sometimes it's harder because there is no functionnal environement at the beginning (eg : i have to port a program under windows from linux). In that case, make sure you know the commonpitfalls (of your toolset or in the code) at first and it's not that hard in the end.

You can also try some 'dirty but working' solutions temporarly.
eg1: instead of managing 20 dll with compilation varitions issue, put every h/cpp file inside a single exe.

Best regards,

Pierre

[Bleakwise]

Google and Google some more. Not always forums, but sometimes RTFM is the best advice you can get. Your subconscious is the most powerful problem-solver you have, but it has to be informed.

If I'm stumped for more than a couple days, maybe a forum post. There was a time where I literally spent months working on understanding a problem, it had to do with linear/matrix algebra which I had never taken in HS/College, so I spent about two months off and on pouring over a couple linear algebra books from the library; actually I'm still in the middle of it as we speak.

I've had a slight history from jumping from project to project as well. I try to limit myself to only two creative endeavor at a single time but I've found that after I've put one on hold, when I come back to it it's not as complex or mysterious as it once seemed.

Maybe I'm not the best person to take advice from (unaccomplished), but that's how it goes for me.

[petersvp]

Google and Google some more. Not always forums, but sometimes RTFM is the best advice you can get. Your subconscious is the most powerful problem-solver you have, but it has to be informed.

If I'm stumped for more than a couple days, maybe a forum post. There was a time where I literally spent months working on understanding a problem, it had to do with linear/matrix algebra which I had never taken in HS/College, so I spent about two months off and on pouring over a couple linear algebra books from the library; actually I'm still in the middle of it as we speak.

I've had a slight history from jumping from project to project as well. I try to limit myself to only two creative endeavor at a single time but I've found that after I've put one on hold, when I come back to it it's not as complex or mysterious as it once seemed.

Maybe I'm not the best person to take advice from (unaccomplished), but that's how it goes for me.

I feel like YOU are ME (or like I am YOU) However I had to eat some physics books as well