Dealing with timeouts

Programming Questions

oolong

in Programming Questions • 2 years ago

I'm doing stuff with xml, pulling it from a web site which sometimes takes its time to respond. As a result, I frequently get

java.net.ConnectException: Connection timed out: connect

in my code, inevitably followed by a NullPointerException, leading the whole program to fail because the server was a bit slow. How should I be dealing with this? Is it a terrible idea to try and catch the exception, and just run the method again if it throws one? Like, in getNodeId I have

try {
xml = new XMLElement(this, "http://everything2.com/title/"+title+"?displaytype=softlinks");
}
catch(Exception e) {
println("Exception in getNodeId: "+e);
return getNodeId(title);
}

I'm obviously not doing this terribly well, because what I get now is the program still producing Null Pointers, but only crashing when it comes to trying to use them for drawing. I was hoping it would keep trying till they weren't Null any more, so I'm clearly missing something.

It's been years since I worked on anything big or faily enough to really need try/catch, so I feel like I may be forgetting some subtleties and possibly doing the whole thing wrong anyway... can anyone give me any pointers here?

Replies(5)

blindfish

Re: Dealing with timeouts

2 years ago

You should probably use a boolean variable to track whether the XML has loaded and test against this before referencing the data in draw (or elsewhere) to avoid crashes... and set it to true when the data has loaded ;)

I guess the only possible risk of calling your XML-loading method recursively on an exception is if for some reason the data simply isn't available (e.g. a typo in a URL) you'll be stuck in an infinite loop. You could perhaps use a counter and allow a certain number of retries before making it fail politely.

oolong

Re: Re: Dealing with timeouts

2 years ago

Thanks, I've incorporated that, as I knew I'd have to - and also figured out that it was failing not because the thing I was trying doesn't work - it does! - but because I'd managed to break another bit of my program without realising. Oops!

blindfish

Re: Dealing with timeouts

2 years ago

Mm, I'm aware that recursively calling the same function every time it throws an exception risks infinite loops if the exception is never resolved. I wondered if there was any other reason it was a bad idea...

Well it being stuck in a loop would also lock up draw() meaning you can't update the screen; which you might want to do if you want to provide feedback to the user re: what's happening. However it sounds like you've now resolved the issue anyway ;)

PhiLho

Re: Dealing with timeouts

2 years ago

Actually, the loop won't be infinite... We have a similar implementation (to connect to an Oracle database) at work, and that's a bad idea, because if the server isn't responding at all, the program is stuck for a while, and then we get an stack overflow due to too much recursion...
It is better to make a controlled loop, as suggested, looping while not getting the data and a counter hasn't reached a value. If there is no data at the end, just politely inform the user...

Jo Wood

Re: Dealing with timeouts

2 years ago

You can also use Java's Timer and TimerTask classes to do external database/XML related stuff in its own thread without interfering with the draw() loop. For example, if you create a simple class in its own tab:

class MyDatabaseTimer extends TimerTask
{
  public void run() 
  { 
    myXMLMethod();

and then in your setup() method of your sketch include the line

timer.scheduleAtFixedRate(new MyDatabaseTimer(), 0, 1000*60*5);

This would call myXMLMethod() every 5 minutes without interfering with the draw loop. You can stop the timer at any point if you have successfully grabbed everything you want, or can leave it running if your content changes dynamically.

I used this approach when creating BikeGrid, which displays the current, dynamically updated, status of 400 bicycle docking stations in central London.

Top Reply