Useful wireshark startup options

Sometime when looking at TCP/IP traffic on multiple ports it's useful to start up several wireshark instances and have each capture traffic on distinct port, and give the window a specific title:


sudo wireshark -o "gui.window_title:Data Server on $SERVER_PORT -" -i lo -p -f "port $SERVER_PORT" -k &


-i lo
capture packets on local loopback interface

-p
don't go into promiscuous mode

-f "port $SERVER_PORT"
use the following filter (specific port)

-k
start capture immediately

How to print process environment variables

Here is a script that prints process environment variables given process ID in Linux:


#!/bin/bash
if [ $# -ne 1 ] ; then
echo "Usage: `basename $0` process_id"
echo "Prints process environment variables"
exit 1
fi

sed 's/\o0/\n/g' /proc/$1/environ


Comparison of Python vs Java

Here is my experience using Python and Java for software development.

Overall, Python is best for small projects and rapid prototyping. Python has certain advantages with its dynamic nature, but as the code base grows these advantages become disadvantages. Java is best for larger projects and where performance is important. Python performance is poor: bytecode executes slow, and global interpreter lock prevents threads from completely utilizing multiple CPUs.

Python is pleasant to use with its expressive string operations, e.g. you can write 'a'*80 to make a string of 80 'a' characters. Using operators on collections in Python is so much more intuitive than using Java's method-based operations. It's nice to not have to compile files.

However, in Python you can make a typo, and it can be frustrating to catch those. Something that background compilation would catch right away in a Java IDE and highlight as syntax error in 1 second, takes writing unit-tests, getting back to the code you've written not at this minute, and fixing it. Because Python is so permissive you can easily do this:

  var = method
  for i in var:
    do something else

Note that you intended to write:

  var = method()  

You omitted the braces. Nothing flagged this as a possible error. During execution, the error isn't manifested on the same line, but instead a couple lines after that when we are trying to do
something with a method instead of an object.

Overall it seems that dynamic languages have inherent limitations that limit code discovery. From a practical point of view, modern languages now have huge libraries, and it's impossible for a human to remember even a fraction of library routine names and parameters. So here is where "suggest" or autocomplete functionality comes in nicely. However it's hard to implement in a dynamic language for the simple reason, that you don't know what types objects are. This impacts your productivity severely, as you have to go and check man pages over and over as opposed to having method names pop up as you write code.

In Python writing performant code is hard. Organizing your code into methods has significant performance penalty if these methods are frequently invoked. Python VM doesn't inline. Essentially Python performance trick is making sure you delegate to the C code. Doing work by executing Python code makes things slow. Because of GIL, Python threads don't truly run simultaneously. Instead it is simulated that they run simultaneously. So if you have multiple CPUs and multiple threads you won't utilize all your CPUs.

Click here for the table summarizing Python and Java.
For some reason Blogger puts a large space between the text and the table.

FEATURE	PYTHON	JAVA
Rapid prototyping	Yes	Less so (compile phase)
Suitable for large projects	Less so hard to refactor (dynamic language) unreliable debugger VM crashes	Yes refactoring support good debugging support
Scalability	Limited (runs only one thread at a time because of GIL)	Better threads run simultaneously
Performance	Slower VM doesn't inline methods Performance penalty for method calls.	Faster Straight bytecode execution about 3 times faster than Python.
Language features	Overall better collection operators e.g. brackets expressive string operations unchecked exceptions non-default constr.inheritance default arguments	Overall worse collections use via methods pluses over python: method overloading
Productivity: Can IDE catch typos?	No Typos become runtime errors Most of the time IDE cannot tell an error from valid use	Yes IDE can compile code in background and highlight typos.
Productivity: IDE: Refactoring, Suggestions, Autocomplete, Code analysis	Hardly possible because Python is dynamic	Yes This makes one write code fast. IDE can perform mathematically correct refactoring
Debugging PyDev vs Eclipse	Rudimental Hard to debug multi-threaded programs: cannot pause all threads Cannot break on exception Unreliable	Superior
Business-friendly	Less so Fewer options as many 3d party libraries tend to be GPL. You may have no option but open sourcing your work. Harder to close source, as normally open py files are distributed instead of compiled bytecode.	More so Many 3d party libraries under BSD, Apache, or LGPL licenses.