Month: January 2020

Java Mission Control / Java Flight Recorder

All

Intro

Sometimes JConsole is not enough to investigate issues, such as performance-related ones.

Java Mission Control 8

Java Mission Control is available for OpenJDK, and Oracle JDK, Oracle has a good introduction for those tools here. Basically, Misson control is the tool for troubleshooting, and Flight Recorder is the API ~ according to JEPS 328.

    Oracle guide to understanding data

Beginning

  1. To install it, go directly from the git: https://github.com/openjdk/jmc
  2. Follow the instruction to build it
  3. Running it

Using JMC

jmc

1. Use case 1:: Analyzing data with Missiong control

1. Use case 2:: JMX to JBoss  (like I did with JVisualVM)

 

Python exercises 2

All

Problem statement

A virus is spreading. It spreads one in all directions. Every hour, it spreads one space in all directions. Given the grid, how many hours it takes it to contaminate the whole grid?

So if you have a grid-like this:

0 0 0 0 1

0 0 0 0 0

0 0 0 0 0

0 0 0 0 0

In one hour it will be like this

0 0 0 1 1

0 0 0 0 1

0 0 0 0 0

0 0 0 0 0

Taking into consideration that the virus can appear anywhere in a grid, which has dimensions that are given by rows x columns, and the grid, which contains the initial virus locals. The virus cell is represented as 1 whereas, normal cells are 0 on the grid.

Resolution

I thought about this problem in two ways: one is procedural, the second is the virus as a living being.

1 – Procedural implementation:

So, you go line by line, get the positions of the virus, and calculate the four next positions ~ validate the positions and change from 0 to 1. In a loop. And continue doing that until all the 0 become 1’s

while(still_zero(grid)):
      print_grid(grid)
      list_ones = find_ones(grid, rows, columns)
      update_nearby(list_ones, grid, rows, columns)
      hours = hours + 1
return hours

 

2 – Node implementation

The virus is actually a node, that grows in four dimensions (so the node knows where it will grow). So, if you have a mechanism to control those nodes, that control this growth. I called a tree but is much more like a linked list of nodes, that know where they will grow.

It will continue to spread while the grid still has 0’s, which is done on the growth process:

while(tree.still_can_grow()):
    aux.print_grid(grid)
    hours = hours + 1
    return hours

Implementation

For the procedural implementation, it is here

The tree implementation is here

Benchmarks

I’m doing some benchmarks on this to see the performance.

 

JSP default buffer size

All

Page buffer

By default, a JSP page uses an area of memory known as a page buffer which is set as 8 KB by default, see ref.

And it is required if the page uses dynamic globalization support content type settings, forwards, or error pages.

You can ad in the page directive the attribute “buffer” like this to change this buffer:

~~~
<%@ page buffer="64kb" ... %>
~~~

And actually, you can disable it with using:

~~~
<%@ page buffer="none" %>
~~~