Write to file

Question

I am having a problem writing to files.
The problem is that I do not get the same result when running on windows xp and windows server 2003 as I do when I run it on a windows server 2008 machine.
The original code was from a larger program but after having debugged this for a day I decided to make a test-program.
My objective: To read a file, after having read 200 lines write lines to new file.
My result: On windows xp and 2003 machines I get 200 lines in new file. On windows 2008 machines I get 218 lines in new file.
Below is a simple test which for me behaves differently on win xp and 2008.
What am I doing wrong?
Code:
private void ProcessFile(string inputFileName, string outputFileName)
{
   Encoding encoding = Encoding.GetEncoding(65001);
   List<string> transactionlist = new List<string>();
   StreamReader sr = new StreamReader(inputFileName, encoding);
   List<string> splittedFiles = new List<string>();
   int lineCount = 0;
   int maxLinesFile = 200;
   while (sr.Peek() != -1)
   {
      lineCount++;
      transactionlist.Add(sr.ReadLine());
      if (lineCount == maxLinesFile)
         break;
   }
   sr.Close();
   sr.Dispose();
   File.WriteAllLines(outputFileName, transactionlist.ToArray(), encoding);
   return;
}

Niklas Buhu · Answer

Thank you all for you responses. I am sad to say I had to give up and install windows 2003 instead on the machine. Then the program works fine. I must say though that I am perplexed what is the reason I can't get it to run on windows server 2008. I have a final thought: I tried to take the code from NLog (a logger library available on the net) because when loggin with this I can write 200 separete log-lines fine without problems. I created my filestream like they did did everything ekstra carefully and still the same result. So I was thinking is it possible it is not the target machines that is at fault but that it is something going wrong when I build it on the developer machine? (But I've tried changing target from 3.5 to 2.0 without a difference to the result). Anyways thank you all for your help. There is still one Windows server 2008 machine I can test on if anyone has a good suggestion.

Niklas Buhu · Answer

Thank you for the suggestion but I've even tried just declaring a string with a predetermined value write it out 200 times while appending a line-number so I can keep track of it. Even then it will repeat itself writing out more than 200 lines. the amount of ekstra text and from which line it will start repeating depends on the length of the line be written out.

PJ Martins · Answer

There's one other thing you could try to do instead of reading the file as text read it as bytes (sr.Read() or File.ReadAllBytes()) and check each individual byte for a new line character. Use a counter variable to keep track of the number of new line characters and once you've hit the 100th newline character stop writing.

Niklas Buhu · Answer

I use UltraEdit to view the files that way I can view linebreaks also. I am now convinced maybe something is wrong with both the windows server 2008 machines I have tested this on even though the exact same thing happens on those 2. I'll explain. If I write out static text flush after each write I get the same error. The error is the same even when I change buffer size. Same error despite using different methods write through File FilestreamStreamwriter or even filestream from kernel32.createfile. I have tried writealllines() writeline and write stating different encodings and buffersizes. I converted the strings to char[] to byte[] using write(byte[] 0 byte[].length) and still same same. So in short I can recreate this using any text using any methods. That it happens on 2 different machines just throws me completely off track :S But I am pretty sure my code is correct (thank you also to PJ martins for checking a earlier version of my test) and this is just too obvious to be bug so only thing I can figure out is that my .net framework must be messed up. I've tried reinstalling through add/remove features but no difference. Have also tried sfc / scannow. Anyone with any suggestions or anyone who is able to test this on a windows server 2008 to confirm that this error must be specific to just my installations?

Sam Hobbs · Answer

You did the right thing by writing a simple sample that recreates the problem; that helps. Instead of using a text editor such as notepad I would compare the files using a binary viewer. In fact Visual Studio is a better text editor than notepad. Using VS you can open a file using either a binary or text editor. Open the two files in VS using the binary editor and use a split window to compare them.

PJ Martins · Answer

Unfortunately I do not have access to a windows server 2008 environment your code does looks fine though one thing you might want to try is to take out the encoding and see if that solves the problem second thing you would have to try is to debug on the windows 2003 server step through the first 99 and then see what happens when the 100th line gets written that causes the extra lines. Sorry I could not be more helpful.

PJ Martins · Answer

Can you attach the file you are trying to copy?

Niklas Buhu · Answer

I tried it just now. Unfortunatly I got the exact same result. 200 lines on windows xp 218 lines when run on windows 2008. I need the encoding as it reads files from various countries it decides the encoding depending on the origin of the file. In this debug though I have just decided on the encoding (65001); Here is the code I tried with; Encoding encoding = Encoding .GetEncoding(65001); StreamReader sr = new StreamReader (inputFileName encoding); StreamWriter sw = new StreamWriter (outputFileName false  encoding 1024); int lineCount = 0; int maxLinesFile = 200; while (sr.Peek() != -1 && lineCount { lineCount++; sw.WriteLine(sr.ReadLine()); } sr.Close(); sr.Dispose(); sw.Close(); sw.Dispose(); I tried changing buffer size but it made no difference. Thank you for your quick response I hope you can help me further also.

PJ Martins · Answer

Any reason you're using that specific encoding? Instead of storing the lines into a list try writing them on the fly using StreamWriter as follows: StreamReader sr = new StreamReader (inputFileName encoding); StreamWriter sw = new StreamWriter (outputFileName encoding); int lineCount = 0; int maxLinesFile = 200; while (sr.Peek() != -1 && lineCount { lineCount++; sw.WriteLine(sr.ReadLine()); } sr.Close(); sr.Dispose(); sw.Close(); sw.Dispose();

Niklas Buhu · Answer

Thats the funny thing. The ekstra 18 lines are really 17½ lines. It repeats the content from line 91 and a half. The files are similar in lengths of each line. I was thinking this might be a buffer problem but I am experiencing the same scenario every time and on 2 different machines. Thank you for answering I was afraid not to get any responses at all and whatever I am doing wrong I can't figure it out.

PJ Martins · Answer

Have you checked the file contents in an editor such as notepad to see if the 18 additional lines are in fact lines from your input file and not something else that got written?