Partnumbers And Other Ideas

This page has been spammed and needs repair

Products have PartNumbers. They are used to identify the part and the number is a construct of many values.

Message Analysis

We measure signalling using programs like wireshark.

We get a printout of each message seen. These are complex with many fields and many protocols.

An expert is expected to understand and report errors and problems.

Often it is difficult to remember all the protocols.

Can we run some test scripts to check the messages - YES, but what do you do. There are many options.

We can compare the messages against a set of templates and examples.

Soon it is appreciated that the order of messages is not always the same and file compare needs to be more sophisticated.

The structure is

• messages
  • mandatory fixed parameters over multiple lines. Is this a tuple.
    • line of multiline value
  • mandatory variable size multiline parameters using Name, length, multiline value structure
    • line 1 of multiline value
    • line 2 of multiline value
    • line 3 of multiline value
  • mandatory variable size multiline parameters using Name, length, multiline value structure
    • line 1 of multiline value
    • line 2 of multiline value
    • ....
    • line n of multiline value

• Optional multiline parameters using Name, length, multiline value structure
  • line 1 of multiline value
  • line 2 of multiline value
  • line 3 of multiline value
• Optional multiline parameters using Name, length, multiline value structure
  • line 1 of multiline value
  • line n of multiline value

Our sort sorts by line, so we need to help it by prepending each line with a tag to be sorted.

The parameters are multiline value structures so lets prepend the line with

  messageName, parameterName , lineNumber : value

This should sort, by message,parameter,line value

Also it groups all values that differ together

Lets write an AWK script that looks for the start of message, start of parameter and build up a sortable string to allow the sort.

http://www.dougrice.plus.com/dev/lines.awk

We also need to work out when to ignore what we are not interested in.

The messages can be:

• fixed length - no optional parameters
• variable length - no optional parameters
• fixed / variable length with optional parameters
• the list of optional parameters may have a NULL pointer to indicate end of list.

These pointers are worth looking for to indicate that the end of the previous parameter has been found.

A test

The capture contains an unordered list of messages. Messages themselves contain a list of fixed parameters, and a list of optional parameters identified by name,length,value, but in no particular order.

We need a compare that accommodates this.

Some parameters may also be an unordered list of name,length,value sub parameters.

We could sort the messages. We could sort the parameters. This is a good idea.

The sort program sorts lines.

So we need to massage the text file so it can be sorted. Many lines will be identicle so we can condense this and count unique lines.

The output is an ordered list of counts of identical lines. This is good as it also shows which values were used and how often they are set to the particular values used.

What do we want from the identical lines?