Problem with writeln

[Ivan]

On 1.3.2010 17:15, Steven Schveighoffer wrote:
> On Mon, 01 Mar 2010 10:42:00 -0500, Ivan <ivan.senji at gmail.com> wrote:
>
>> On 1.3.2010 14:43, Steven Schveighoffer wrote:
>>> On Mon, 01 Mar 2010 07:14:23 -0500, Ivan <ivan.senji at gmail.com> wrote:
>>>
>>>> On 1.3.2010 12:47, Steven Schveighoffer wrote:
>>>>> On Mon, 01 Mar 2010 04:11:28 -0500, Ivan <ivan.senji at gmail.com> wrote:
>>>>>
>>>>>> The part of code in my main method is this:
>>>>>>
>>>>>> try {
>>>>>> writefln("Starting main loop...");
>>>>>> global.loop.start();
>>>>>> writefln("Main loop finished...");
>>>>>> } catch(Exception e) {
>>>>>> writefln("Exception was: %s", e.msg);
>>>>>> }
>>>>>>
>>>>>> From what I can see an exception was thrown ("unable to write to
>>>>>> stream") and the only line I added to the loop is this:
>>>>>>
>>>>>> writefln(". %s", counter++);
>>>>>>
>>>>>> If that line is removed, an exception seams to be thrown from the
>>>>>> line
>>>>>> writefln("Main loop finished...");
>>>>>
>>>>> Post more code. I suspect you are messing up something. It appears
>>>>> that
>>>>> you may be closing stdout, but I can't be sure because your code is
>>>>> not
>>>>> complete.
>>>>>
>>>>
>>>> The code can be seen here: http://github.com/ivans/ifs3D
>>>> I didn't and still don't want to bother anyone too much. :)
>>>>
>>>> It should compile on windows with a command found in a readme file.
>>>>
>>>> I don't think I am closing stdout anywhere and I probably am doing
>>>> something wrong somewhere, but can't figure out where and what.
>>>
>>> Your code is too complex for me to understand quickly :)
>>>
>>> Comment out other lines until you figure out which one closes it.
>>>
>>> Start with commenting out the entire loop :) If that then works, you
>>> know something inside the loop is killing stdout.
>>>
>>> BTW, are you compiling in Windows mode? I.e. do you see a console on the
>>> screen when you run it? If not, then there is no stdout or stdin. From
>>> your command line in the readme, I don't think you are. However, in that
>>> case, I don't think you'd get anywhere or see any output.
>>
>> Thank you for helping. I did what you suggested and started commenting
>> out code to get it working. I finally got to a working code with just
>> two lines commented out:
>>
>> http://github.com/ivans/ifs3D/commit/339ca5ee58661d636bfcd878c08b06bacede7dca
>>
>>
>> These lines have nothing to do with streams and stdout, so I still
>> don't understand what is going on. But this part of code is very old
>> and stupid (allocating a lot of unneccessary real[][]'s) so I guess
>> the next step is to figure out a smarter way to do that part and see
>> if the exception goes away :).
>
> I suspect there may be a bug in dmd when dealing with reals,
> specifically with your array creation routine.
>
> That code doesn't seem to be incorrect by any means.
>
> Try building the resulting arrays manually (that is, instead of using
> that array template function), and see if it helps.
>
> Also, try removing the calls to the inner sq function (BTW, you don't
> need to make that an inner function, doing so unnecessarily complicates
> things).
>
> If you can find the thing that fixes it, we may be able to find a
> reduced test case. Also, try upgrading your compiler to the latest if
> you haven't already.

The problem wasn't with reals. I added an alias so I can easily change 
my floating point type and application crashed with all of them, but 
after a diferent amount of iterations.

I managed to get rid of the problem by removing a lot of dynamic array 
allocations and I am now mostly using static arrays. I still have this 
same problem in another peace of that same program and will still try to 
get a reproducable and small test case.

Thanks for all the help and good advice. :)