listener.d: Range violation.

[Dmitry Olshansky]

On 30.05.2011 0:38, Jesse Brands wrote:
> Hello there,
>
> For a small chat server thingie I'm making, I decided to use D; finding myself
> with a very neat example from listener.d to get a kick start I decided to
> pretty much take the example over! However, I'm stuck on a bug I can't truly
> wrap my finger around. Most likely it's my own fault and I'm doing something
> wrong, but considering I took the code pretty much from the example I am more
> inclined to believe the example is broken.
>
> I'll explain what happens:
>   -- I start up my server (nothing wrong, it's running as it should and listening)
>   -- I telnet to it. My server accepts the connection.
>   -- I use telnet to send some information. Server handles the information
> properly, again, no issue.
>   -- I quit telnet by using ^] and then writing quit. Breaking the connection
> rather ungracefully.
>   -- The server properly recognises this isn't a clean disconnect and executes
> the code to remove the socket.
>   -- I then get a range violation.
>
> This is the main process and it's loop:
> https://github.com/JGBrands/BlaatServer/blob/master/source/bserver.d
>
> This is the server class, the code where it deletes the socket is at the
> complete bottom in the function void destroySocket(int index);
> https://github.com/JGBrands/BlaatServer/blob/master/source/server.d
>
> The code is primarily taken over from the listener.d example like I said, the
> error I get is this:
> core.exception.RangeError at server(61): Range violation
> ----------------
> ----------------
>
> I'm lead to believe the function is deleting something it shouldn't, for those
> interested, this is line 61 in server.d:
> if (this.sset.isSet(this.reads[i])) {
Just a stub in the dark, but since you are killing the last one socket, 
(that is calling destroySocket in next)
but you get your index still incremented (next returns true) so it will 
point out of array and it won't be equal to reads.length on next step.
Try adjusting that == to >= on line 60 to test this theory.

-- 
Dmitry Olshansky