quickfix/n how to change the encoding from iso-8859-1 to windows-1251

[githubniko]

Hello. The server transmits FIX messages in windows-1251 encoding. QuickFix writes data to a log file in iso-8859-1. Is it possible to change the encoding without recompiling quickfix/n ?

[mgatny]

Are you using an older version of qf/n?

I think we very explicity transmit in iso-8859-1 (latin1) which is the protocol standard:

~/dev/quickfixn/QuickFIXn$ cat CharEncoding.cs; grep -A5 " Send(" SocketReader.cs SocketInitiatorThread.cs 
namespace QuickFix
{
    public static class CharEncoding
    {
        public static System.Text.Encoding DefaultEncoding;

        static CharEncoding()
        {
            DefaultEncoding = System.Text.Encoding.GetEncoding("iso-8859-1");
        }
    }
}
SocketReader.cs:        public int Send(string data)
SocketReader.cs-        {
SocketReader.cs-            byte[] rawData = CharEncoding.DefaultEncoding.GetBytes(data);
SocketReader.cs-            stream_.Write(rawData, 0, rawData.Length);
SocketReader.cs-            return rawData.Length;
SocketReader.cs-        }
--
SocketInitiatorThread.cs:        public bool Send(string data)
SocketInitiatorThread.cs-        {
SocketInitiatorThread.cs-            byte[] rawData = CharEncoding.DefaultEncoding.GetBytes(data);
SocketInitiatorThread.cs-            stream_.Write(rawData, 0, rawData.Length);
SocketInitiatorThread.cs-            return true;
SocketInitiatorThread.cs-        }

[githubniko]

I want the encoding to be taken out to the settings. This will make quick fix more flexible.

"ISO-8859-1" is a standard for English-speaking people. My broker uses win1251 by default and I get a crooked text display and then I have to recode the iso-8859-1 - >win1251, which consumes computing resources.

Now I have edited win1251 encoding in the Char Encoding.cs file and everything has worked for me. Example:

public static class CharEncoding
    {
        public static System.Text.Encoding Default Encoding;

        static Char Encoding()
        {
            ////Default Encoding = System.Text.Encoding.GetEncoding("iso-8859-1");
            Default Encoding = System.Text.Encoding.GetEncoding(1251);
        }
    }

1. Default Encoding = System.Text.Encoding.GetEncoding("iso-8859-1");
2. (right) Default Encoding = System.Text.Encoding.GetEncoding(1251);

[mgatny]

I wouldn't be opposed to exposing it as a SessionSetting. Although the FIX Protocol standard requires latin1, many firms (including very large ones) obviously do not adhere to it.

[jded76]

We faced the same problem with the encoding. Is there any chance to implement this?

There was a commit (https://github.com/connamara/quickfixn/commit/e7d59cea8e46c4f065d4834a4acd171407ea220f)) that implemented that, but this commit (https://github.com/connamara/quickfixn/commit/a75fc7caf0adaf513fabd4aa3378c10797cc7494) reverted back.

Thank you

[jded76]

Can I do something to help on this? I think it's an easy fix and without any danger (if you don't change it, it will work as it is right now).

[jded76]

I said an easy fix because I have in mind that I don't need it per session. We can have a similar approach that QuickFixJ has (see here). A static SetEncoding that we can call one time at the beginning. So the fix will be easy enough.

[jded76]

@gbirchmeier I saw that there is activity lately with a lot of changes. Any change to take a look in this too (see my previous comments)?

We have forked and build our own version just to support a different encoding, so we are stuck without updates.

Thank you

P.S. QuickfixJ current implementation here