Libraries for analyze Janpanese text

https://github.com/SamuraiT/mecab-python3
http://mecab.googlecode.com/svn/trunk/mecab/doc/index.html


https://github.com/Christoph-D/Japanese-Tools/

Run many services on same server

If a service goes wrong, using 100% -> server is unresponsive, down!

To address this, I finally found:

cgroups: Control CPU/Mem resources for group of process

https://wiki.archlinux.org/index.php/cgroups

nice, cpulimit: Control CPU utilization level.

coreos + docker -> everything will be under measured.

Open big files in Komodo

Although Komodo is great, it will be broken on big files. The reason is costly syntax highlight.

Big files can be: too many lines; or big filesize.

To overcome this, create and run this macro "once".
After running, it will be persisted as a preference.
My setting is 200Kb and 5000 lines limit : )

// Let Komodo able to open big files by disable Syntax Highlight
var gprefs = Components.classes["@activestate.com/koPrefService;1"].
  getService(Components.interfaces.koIPrefService).prefs;
gprefs.setLongPref("documentByteCountThreshold", 200000);
gprefs.setLongPref("documentLineCountThreshold", 5000);
alert('All was set. Openning big files is OK right now');

Ruby notes

Learn Ruby language via style guide is interesting
https://github.com/bbatsov/ruby-style-guide

Goodies:
http://docs.ruby-lang.org/en/2.2.0/Enumerable.html
http://docs.ruby-lang.org/en/2.2.0/syntax/modules_and_classes_rdoc.html

My discoveries:

print : echo bare text

p: print object as it as, with new line

puts: echo text with new line

alias create alias for existing method, var

ensure always executed at block termination

block comment

=begin 

=end

undef remove function definition

convention:

Eg: eql?

method with a question mark return a bool value

Eg: delete! (bang symbol)

method with exclamation ! indicates that it's destructive: change object than a copy

Eg: name=

method ends with = is a setter

block can be wrapped by "do .. end" or { }

run by "yield"

Check for existence via if block_given?

:symbol 

storage optimized string

Global names is from Kernel module

replace string once:

string['search']= 'replace'

replace all

string.gsub 'search', 'replace'

regex

string[/regex/]

regex replace uses the same syntax

string.split(/, /) you might guess

Object methods:

obj.to_s

obj.class

string.to_sym -> :symbol

num_string.to_i -> integer

range.to_a -> array

%w[ string1, ... ]

%i [ symbol1, ...]

[num1, ... ]

array << x # Add x to array (immuttable)

set operator: intersect &, diffeence -, union | 

array.join sep

array.shift

Hash.new(default_value)

hash.key?

hash.value?

hash.keys

hash.values

Gem gotcha

If your local machine's hostname is set to your name, you will likely hit a trouble while installing gem:

"your-dns-needs-immediate-attention"

That's all I know. That message is oversay!

What to do is: set your hostname to a real top-level domain.
If it's a local machine, set it to "your-name.local"
Bold characters must be exactly typed!

# Linux

hostname -s your-name.local 

# Mac OSX

scutil --set HostName your-name.local

Dreamy app stack

If choosing my own stack, I will choose:

- Python: Beauty, simple, powerful. It's slow but productive. Anything has tradeoff ; )
Framework: Tornado

- Cassandra: Autoscale, SQL-like query, Fast write, able to tune consistency: What can I ask more for a database!

- WeedFS: Distribute storage.

Find PHP bottleneck!

There are many ways, such as Xhprof. However, because I already had Xdebug I will use Xdebug to profile the app; and setup Xdebug seems easier than Xhprof.

1. First set these in .htaccess (or php.ini)

php_value xdebug.profiler_enable 1
php_value xdebug.profiler_output_dir "/var/xdebug/"

2. Make sure that /var/xdebug/ is writable by web app, or change it to your folder.

3. Run the request to benchmark.

4. Each request will result in a new file /var/xdebug/cachegrind.out.NNNN

5. To read these stats needs brew install qcachegrind on Mac OSX, or kcachegrind on Linux.
After installing them, run, eg: qcachegrind /var/xdebug/cachegrind.out.NNNN

6. Examine by clicking on "self" column; consider function that take longest time. Well from here we need to think about what caused these bottlenecks.