Related Plugins and Tags

QGIS Planet

Generating state machines with Dia

Dia is a nice cross platform application for diagram drawing. It can be scripted via Python, which opens the possibility to generate code from Dia diagrams.

Below we’ll create a python plugin that generates C code from UML state machine diagrams. Doing the same for other languages should be trivial.

The first thing you need is Unai Estébanez Sevilla’s nice finite state machine code generator. The version that we are using here has been abstracted in order to be able to produce code in various languages.

This python plugin implements an exporter to C code. To use it, you need to put it, along with the base exporter into your local plugins directory under ~/.dia/python.

Let’s have a quick look at the code.

First we import the dia python module and the exporter base functionality:

import dia
import uml_stm_export

Then we create our C exporter class that inherits from the generic exporter:

class CDiagramRenderer(uml_stm_export.SimpleSTM):

Next we define how the beginning of our generated code file should look like. That could include general infrastructure independent of the state machine diagram at hand. In our case, we want to encapsulate the generated state machine code within a function:

CODE_PREAMBLE="void config_stm(STM_t* stm) {"

We also define the postamble to close the function. After that come generic functions that implement the class constructor init(self) and functions responsible for calling the dia object parser begin_render(self,data,filename).

Now we define our output generator end_render(self). We first traverse dia’s objects in order to find the state machine’s initial state:

for transition in self.transitions:
   if(transition.source == "INITIAL_STATE"):

The initial state state gets a special treatment: we have a special function call generated for it:

f.write("    add_initial_state( stm, %s, %s );\n" %
    (initial_state.name, initial_state.doaction))

Next we traverse all states and output code that will create them, along with functions to be called within that state to decide on where to transition next:

for key in self.states.keys():
    f.write("    add_state( stm, %s, %s );\n"
        % (state.name, state.doaction))

And finally we output all the transitions between states:

for transition in self.transitions:
    f.write("    add_transition( stm, %s, %s, %s );\n" %
        (transition.source, transition.trigger, transition.target))

and that’s nearly it. At the end of our generator we make sure to register it with dia:

dia.register_export("State Machine Cstma Dump", "c", CDiagramRenderer())

Done! Simple, isn’t it?

Finally please permit me to thank all the people that created such a powerful tool free for us to use:

  • Unai Estébanez Sevilla for the original STM generator
  • Steffen Macke and Hans Breuer, Dia’s current busy maintaners
  • Alexander Larsson, Dia’s original author
  • all the other contributors to Dia and free software
  • Panter for inviting me to their fabulous work week in Greece where most of the hacking on the generator was done and Combitool who supported this work by needing a state machine generator in their current project.

PS: Unai’s original text generator is now also “just” a “simple” addon

Generating state machines with Dia

Dia is a nice cross platform application for diagram drawing. It can be scripted via Python, which opens the possibility to generate code from Dia diagrams.

Below we’ll create a python plugin that generates C code from UML state machine diagrams. Doing the same for other languages should be trivial.

The first thing you need is Unai Estébanez Sevilla’s nice finite state machine code generator. The version that we are using here has been abstracted in order to be able to produce code in various languages.

This python plugin implements an exporter to C code. To use it, you need to put it, along with the base exporter into your local plugins directory under ~/.dia/python.

Let’s have a quick look at the code.

First we import the dia python module and the exporter base functionality:

import dia
import uml_stm_export

Then we create our C exporter class that inherits from the generic exporter:

class CDiagramRenderer(uml_stm_export.SimpleSTM):

Next we define how the beginning of our generated code file should look like. That could include general infrastructure independent of the state machine diagram at hand. In our case, we want to encapsulate the generated state machine code within a function:

CODE_PREAMBLE="void config_stm(STM_t* stm) {"

We also define the postamble to close the function. After that come generic functions that implement the class constructor init(self) and functions responsible for calling the dia object parser begin_render(self,data,filename).

Now we define our output generator end_render(self). We first traverse dia’s objects in order to find the state machine’s initial state:

for transition in self.transitions:
   if(transition.source == "INITIAL_STATE"):

The initial state state gets a special treatment: we have a special function call generated for it:

f.write("    add_initial_state( stm, %s, %s );\n" %
    (initial_state.name, initial_state.doaction))

Next we traverse all states and output code that will create them, along with functions to be called within that state to decide on where to transition next:

for key in self.states.keys():
    f.write("    add_state( stm, %s, %s );\n"
        % (state.name, state.doaction))

And finally we output all the transitions between states:

for transition in self.transitions:
    f.write("    add_transition( stm, %s, %s, %s );\n" %
        (transition.source, transition.trigger, transition.target))

and that’s nearly it. At the end of our generator we make sure to register it with dia:

dia.register_export("State Machine Cstma Dump", "c", CDiagramRenderer())

Done! Simple, isn’t it?

Finally please permit me to thank all the people that created such a powerful tool free for us to use:

  • Unai Estébanez Sevilla for the original STM generator
  • Steffen Macke and Hans Breuer, Dia’s current busy maintaners
  • Alexander Larsson, Dia’s original author
  • all the other contributors to Dia and free software
  • Panter for inviting me to their fabulous work week in Greece where most of the hacking on the generator was done and Combitool who supported this work by needing a state machine generator in their current project.

PS: Unai’s original text generator is now also “just” a “simple” addon

Reducing magic in Ruby's Forwardable class implementation

Abstract

Using string-eval in Ruby for metaprogramming is unnecessarily obscuring. Ruby’s more modern and specific metaprogramming methods should be used instead whenever possible. This problem is illustrated on the example of Ruby’s Forwardable class.

In detail…

Ruby’s Forwardable class is using metaprogramming to forward calls from a frontend interface to an instance in the back executing the call.

Metaprogramming is the discipline of making code that creates code. This task allready is rather abstract and hard to grasp in itself. Having hard to grasp code is a liability. One of the goals of writing code is allways to keep the code as simple and as well understandable as possible.

Additionaly, metaprogramming code itself is difficult to read and understand: that is because the metaprogramming code will not necessarily express what the code it is creating is about, but only how it is creating that code. As such the code it is creating can be invisible to you as a reader of the source code - the created code will only start to exist at runtime.

One would therefore expect that programmers would try especially hard when they metaprogram to make that particular kind of code expressive and easy to understand.

Another consequence of the fact that the code produced by metaprogramming is not necessarily visible, is that debugging becomes more difficult: when analyzing problems you’ll not only be unsure how the programm works, but in addition, you won’t even be sure how the code that is executed looks like - since it is only generated at runtime.

This post is focusing on the last problem: debugging of metaprogrammed code.

There are two approaches to metaprogramming. One is to have as far as possible compile-time parseable code and the other is to let the code only be parsed at runtime.

As of version 1.9.2, Ruby’s Forwardable class is using the latter. The metaprogramming code in Ruby 1.8.7 looks like this:

    module_eval(<

As said, this has the consequence of the metaprogrammed code being completely invisible to the parser and other tools such as editors and debuggers.

This results in the following:

$ cat queue.rb
require 'rubygems'
require 'forwardable'
require 'ruby-debug'

class Queue
  extend Forwardable

  def initialize
    @q = [ ]    # prepare delegate object
  end

  # setup preferred interface, enq() and deq()...
  def_delegator :@q, :push, :enq
  def_delegator :@q, :shift, :deq

  # support some general Array methods that fit Queues well
  def_delegators :@q, :clear, :first, :push, :shift, :size
end

q = Queue.new
debugger # ------ DEBUGGING FROM HERE ON -----
q.enq 1, 2, 3, 4, 5
q.push 6

q.shift    # => 1
while q.size > 0
  puts q.deq
end

q.enq "Ruby", "Perl", "Python"
puts q.first
q.clear
puts q.first


$ ruby queue.rb

queue.rb:24
q.enq 1, 2, 3, 4, 5

(rdb:1) step
(__FORWARDABLE__):2

(rdb:1) list =
*** No sourcefile available for (__FORWARDABLE__)

(rdb:1) step
(__FORWARDABLE__):3

In other words, you are rather lost allready - otherwise you probably wouldn’t be stepping through your code - and in that situation it happens that your debugger gets completely lost as well, since it does not know any more where in the code it is and what it exactly is executing.

That’s nothing the programmer wishes for. In a situation where you are debugging you want to have a maximally clear view of all state, including what code you are currently executing.

Chaning that situation requires making as much of the metaprogrammed code visible to the parser, which is the second approach to metaprogramming mentioned previously:

$ cat forwardable2.rb
...
    self.send(:define_method, ali) do |*args,█|
      begin
        instance_variable_get(accessor).__send__(method, *args,█)
      rescue Exception
        [email protected]_if{|s| /^\\(__FORWARDABLE__\\):/ =~ s} unless Forwardable2::debug
        Kernel::raise
      end
    end

Note that it’s the same code as before, except that we do not do eval("string") any more, but instead are using specific, more modern metaprogramming tools provided by standard Ruby.

The result is the following:

$ ruby queue.rb

queue.rb:22
q.enq 1, 2, 3, 4, 5

(rdb:1) step
/usr/lib/ruby/1.8/forwardable2.rb:149
begin

(rdb:1) list =
[144, 153] in /usr/lib/ruby/1.8/forwardable2.rb
   144      accessor = accessor.id2name if accessor.kind_of?(Integer)
   145      method = method.id2name if method.kind_of?(Integer)
   146      ali = ali.id2name if ali.kind_of?(Integer)
   147
   148      self.send(:define_method, ali) do |*args,█|
=> 149        begin
   150          instance_variable_get(accessor).__send__(method, *args,█)
   151        rescue Exception
   152          [email protected]_if{|s| /^\\(__FORWARDABLE__\\):/ =~ s} unless Forwardable2::debug
   153          Kernel::raise

(rdb:1) step
/usr/lib/ruby/1.8/forwardable2.rb:150
instance_variable_get(accessor).__send__(method, *args,█)

Allready much, much better.

Of course, with the string-eval approach to metaprogramming Ruby itself could do better by saving the string that is being evaled to be able to refer to it later at step-through time. However currently we don’t have this option.

Tomáš Pospíšek

Reducing magic in Ruby's Forwardable class implementation

Abstract

Using string-eval in Ruby for metaprogramming is unnecessarily obscuring. Ruby’s more modern and specific metaprogramming methods should be used instead whenever possible. This problem is illustrated on the example of Ruby’s Forwardable class.

In detail…

Ruby’s Forwardable class is using metaprogramming to forward calls from a frontend interface to an instance in the back executing the call.

Metaprogramming is the discipline of making code that creates code. This task allready is rather abstract and hard to grasp in itself. Having hard to grasp code is a liability. One of the goals of writing code is allways to keep the code as simple and as well understandable as possible.

Additionaly, metaprogramming code itself is difficult to read and understand: that is because the metaprogramming code will not necessarily express what the code it is creating is about, but only how it is creating that code. As such the code it is creating can be invisible to you as a reader of the source code - the created code will only start to exist at runtime.

One would therefore expect that programmers would try especially hard when they metaprogram to make that particular kind of code expressive and easy to understand.

Another consequence of the fact that the code produced by metaprogramming is not necessarily visible, is that debugging becomes more difficult: when analyzing problems you’ll not only be unsure how the programm works, but in addition, you won’t even be sure how the code that is executed looks like - since it is only generated at runtime.

This post is focusing on the last problem: debugging of metaprogrammed code.

There are two approaches to metaprogramming. One is to have as far as possible compile-time parseable code and the other is to let the code only be parsed at runtime.

As of version 1.9.2, Ruby’s Forwardable class is using the latter. The metaprogramming code in Ruby 1.8.7 looks like this:

    module_eval(<

As said, this has the consequence of the metaprogrammed code being completely invisible to the parser and other tools such as editors and debuggers.

This results in the following:

$ cat queue.rb
require 'rubygems'
require 'forwardable'
require 'ruby-debug'

class Queue
  extend Forwardable

  def initialize
    @q = [ ]    # prepare delegate object
  end

  # setup preferred interface, enq() and deq()...
  def_delegator :@q, :push, :enq
  def_delegator :@q, :shift, :deq

  # support some general Array methods that fit Queues well
  def_delegators :@q, :clear, :first, :push, :shift, :size
end

q = Queue.new
debugger # ------ DEBUGGING FROM HERE ON -----
q.enq 1, 2, 3, 4, 5
q.push 6

q.shift    # => 1
while q.size > 0
  puts q.deq
end

q.enq "Ruby", "Perl", "Python"
puts q.first
q.clear
puts q.first


$ ruby queue.rb

queue.rb:24
q.enq 1, 2, 3, 4, 5

(rdb:1) step
(__FORWARDABLE__):2

(rdb:1) list =
*** No sourcefile available for (__FORWARDABLE__)

(rdb:1) step
(__FORWARDABLE__):3

In other words, you are rather lost allready - otherwise you probably wouldn’t be stepping through your code - and in that situation it happens that your debugger gets completely lost as well, since it does not know any more where in the code it is and what it exactly is executing.

That’s nothing the programmer wishes for. In a situation where you are debugging you want to have a maximally clear view of all state, including what code you are currently executing.

Chaning that situation requires making as much of the metaprogrammed code visible to the parser, which is the second approach to metaprogramming mentioned previously:

$ cat forwardable2.rb
...
    self.send(:define_method, ali) do |*args,█|
      begin
        instance_variable_get(accessor).__send__(method, *args,█)
      rescue Exception
        [email protected]_if{|s| /^\\(__FORWARDABLE__\\):/ =~ s} unless Forwardable2::debug
        Kernel::raise
      end
    end

Note that it’s the same code as before, except that we do not do eval("string") any more, but instead are using specific, more modern metaprogramming tools provided by standard Ruby.

The result is the following:

$ ruby queue.rb

queue.rb:22
q.enq 1, 2, 3, 4, 5

(rdb:1) step
/usr/lib/ruby/1.8/forwardable2.rb:149
begin

(rdb:1) list =
[144, 153] in /usr/lib/ruby/1.8/forwardable2.rb
   144      accessor = accessor.id2name if accessor.kind_of?(Integer)
   145      method = method.id2name if method.kind_of?(Integer)
   146      ali = ali.id2name if ali.kind_of?(Integer)
   147
   148      self.send(:define_method, ali) do |*args,█|
=> 149        begin
   150          instance_variable_get(accessor).__send__(method, *args,█)
   151        rescue Exception
   152          [email protected]_if{|s| /^\\(__FORWARDABLE__\\):/ =~ s} unless Forwardable2::debug
   153          Kernel::raise

(rdb:1) step
/usr/lib/ruby/1.8/forwardable2.rb:150
instance_variable_get(accessor).__send__(method, *args,█)

Allready much, much better.

Of course, with the string-eval approach to metaprogramming Ruby itself could do better by saving the string that is being evaled to be able to refer to it later at step-through time. However currently we don’t have this option.

Tomáš Pospíšek

Reducing magic in Ruby's Forwardable class implementation

Abstract

Using string-eval in Ruby for metaprogramming is unnecessarily obscuring. Ruby’s more modern and specific metaprogramming methods should be used instead whenever possible. This problem is illustrated on the example of Ruby’s Forwardable class.

In detail…

Ruby’s Forwardable class is using metaprogramming to forward calls from a frontend interface to an instance in the back executing the call.

Metaprogramming is the discipline of making code that creates code. This task allready is rather abstract and hard to grasp in itself. Having hard to grasp code is a liability. One of the goals of writing code is allways to keep the code as simple and as well understandable as possible.

Additionaly, metaprogramming code itself is difficult to read and understand: that is because the metaprogramming code will not necessarily express what the code it is creating is about, but only how it is creating that code. As such the code it is creating can be invisible to you as a reader of the source code - the created code will only start to exist at runtime.

One would therefore expect that programmers would try especially hard when they metaprogram to make that particular kind of code expressive and easy to understand.

Another consequence of the fact that the code produced by metaprogramming is not necessarily visible, is that debugging becomes more difficult: when analyzing problems you’ll not only be unsure how the programm works, but in addition, you won’t even be sure how the code that is executed looks like - since it is only generated at runtime.

This post is focusing on the last problem: debugging of metaprogrammed code.

There are two approaches to metaprogramming. One is to have as far as possible compile-time parseable code and the other is to let the code only be parsed at runtime.

As of version 1.9.2, Ruby’s Forwardable class is using the latter. The metaprogramming code in Ruby 1.8.7 looks like this:

    module_eval(<

As said, this has the consequence of the metaprogrammed code being completely invisible to the parser and other tools such as editors and debuggers.

This results in the following:

$ cat queue.rb
require 'rubygems'
require 'forwardable'
require 'ruby-debug'

class Queue
  extend Forwardable

  def initialize
    @q = [ ]    # prepare delegate object
  end

  # setup preferred interface, enq() and deq()...
  def_delegator :@q, :push, :enq
  def_delegator :@q, :shift, :deq

  # support some general Array methods that fit Queues well
  def_delegators :@q, :clear, :first, :push, :shift, :size
end

q = Queue.new
debugger # ------ DEBUGGING FROM HERE ON -----
q.enq 1, 2, 3, 4, 5
q.push 6

q.shift    # => 1
while q.size > 0
  puts q.deq
end

q.enq "Ruby", "Perl", "Python"
puts q.first
q.clear
puts q.first


$ ruby queue.rb

queue.rb:24
q.enq 1, 2, 3, 4, 5

(rdb:1) step
(__FORWARDABLE__):2

(rdb:1) list =
*** No sourcefile available for (__FORWARDABLE__)

(rdb:1) step
(__FORWARDABLE__):3

In other words, you are rather lost allready - otherwise you probably wouldn’t be stepping through your code - and in that situation it happens that your debugger gets completely lost as well, since it does not know any more where in the code it is and what it exactly is executing.

That’s nothing the programmer wishes for. In a situation where you are debugging you want to have a maximally clear view of all state, including what code you are currently executing.

Chaning that situation requires making as much of the metaprogrammed code visible to the parser, which is the second approach to metaprogramming mentioned previously:

$ cat forwardable2.rb
...
    self.send(:define_method, ali) do |*args,█|
      begin
        instance_variable_get(accessor).__send__(method, *args,█)
      rescue Exception
        [email protected]_if{|s| /^\\(__FORWARDABLE__\\):/ =~ s} unless Forwardable2::debug
        Kernel::raise
      end
    end

Note that it’s the same code as before, except that we do not do eval("string") any more, but instead are using specific, more modern metaprogramming tools provided by standard Ruby.

The result is the following:

$ ruby queue.rb

queue.rb:22
q.enq 1, 2, 3, 4, 5

(rdb:1) step
/usr/lib/ruby/1.8/forwardable2.rb:149
begin

(rdb:1) list =
[144, 153] in /usr/lib/ruby/1.8/forwardable2.rb
   144      accessor = accessor.id2name if accessor.kind_of?(Integer)
   145      method = method.id2name if method.kind_of?(Integer)
   146      ali = ali.id2name if ali.kind_of?(Integer)
   147
   148      self.send(:define_method, ali) do |*args,█|
=> 149        begin
   150          instance_variable_get(accessor).__send__(method, *args,█)
   151        rescue Exception
   152          [email protected]_if{|s| /^\\(__FORWARDABLE__\\):/ =~ s} unless Forwardable2::debug
   153          Kernel::raise

(rdb:1) step
/usr/lib/ruby/1.8/forwardable2.rb:150
instance_variable_get(accessor).__send__(method, *args,█)

Allready much, much better.

Of course, with the string-eval approach to metaprogramming Ruby itself could do better by saving the string that is being evaled to be able to refer to it later at step-through time. However currently we don’t have this option.

Tomáš Pospíšek

Reducing magic in Ruby's Forwardable class implementation

Abstract

Using string-eval in Ruby for metaprogramming is unnecessarily obscuring. Ruby’s more modern and specific metaprogramming methods should be used instead whenever possible. This problem is illustrated on the example of Ruby’s Forwardable class.

In detail…

Ruby’s Forwardable class is using metaprogramming to forward calls from a frontend interface to an instance in the back executing the call.

Metaprogramming is the discipline of making code that creates code. This task allready is rather abstract and hard to grasp in itself. Having hard to grasp code is a liability. One of the goals of writing code is allways to keep the code as simple and as well understandable as possible.

Additionaly, metaprogramming code itself is difficult to read and understand: that is because the metaprogramming code will not necessarily express what the code it is creating is about, but only how it is creating that code. As such the code it is creating can be invisible to you as a reader of the source code - the created code will only start to exist at runtime.

One would therefore expect that programmers would try especially hard when they metaprogram to make that particular kind of code expressive and easy to understand.

Another consequence of the fact that the code produced by metaprogramming is not necessarily visible, is that debugging becomes more difficult: when analyzing problems you’ll not only be unsure how the programm works, but in addition, you won’t even be sure how the code that is executed looks like - since it is only generated at runtime.

This post is focusing on the last problem: debugging of metaprogrammed code.

There are two approaches to metaprogramming. One is to have as far as possible compile-time parseable code and the other is to let the code only be parsed at runtime.

As of version 1.9.2, Ruby’s Forwardable class is using the latter. The metaprogramming code in Ruby 1.8.7 looks like this:

    module_eval(<

As said, this has the consequence of the metaprogrammed code being completely invisible to the parser and other tools such as editors and debuggers.

This results in the following:

$ cat queue.rb
require 'rubygems'
require 'forwardable'
require 'ruby-debug'

class Queue
  extend Forwardable

  def initialize
    @q = [ ]    # prepare delegate object
  end

  # setup preferred interface, enq() and deq()...
  def_delegator :@q, :push, :enq
  def_delegator :@q, :shift, :deq

  # support some general Array methods that fit Queues well
  def_delegators :@q, :clear, :first, :push, :shift, :size
end

q = Queue.new
debugger # ------ DEBUGGING FROM HERE ON -----
q.enq 1, 2, 3, 4, 5
q.push 6

q.shift    # => 1
while q.size > 0
  puts q.deq
end

q.enq "Ruby", "Perl", "Python"
puts q.first
q.clear
puts q.first


$ ruby queue.rb

queue.rb:24
q.enq 1, 2, 3, 4, 5

(rdb:1) step
(__FORWARDABLE__):2

(rdb:1) list =
*** No sourcefile available for (__FORWARDABLE__)

(rdb:1) step
(__FORWARDABLE__):3

In other words, you are rather lost allready - otherwise you probably wouldn’t be stepping through your code - and in that situation it happens that your debugger gets completely lost as well, since it does not know any more where in the code it is and what it exactly is executing.

That’s nothing the programmer wishes for. In a situation where you are debugging you want to have a maximally clear view of all state, including what code you are currently executing.

Chaning that situation requires making as much of the metaprogrammed code visible to the parser, which is the second approach to metaprogramming mentioned previously:

$ cat forwardable2.rb
...
    self.send(:define_method, ali) do |*args,█|
      begin
        instance_variable_get(accessor).__send__(method, *args,█)
      rescue Exception
        [email protected]_if{|s| /^\\(__FORWARDABLE__\\):/ =~ s} unless Forwardable2::debug
        Kernel::raise
      end
    end

Note that it’s the same code as before, except that we do not do eval("string") any more, but instead are using specific, more modern metaprogramming tools provided by standard Ruby.

The result is the following:

$ ruby queue.rb

queue.rb:22
q.enq 1, 2, 3, 4, 5

(rdb:1) step
/usr/lib/ruby/1.8/forwardable2.rb:149
begin

(rdb:1) list =
[144, 153] in /usr/lib/ruby/1.8/forwardable2.rb
   144      accessor = accessor.id2name if accessor.kind_of?(Integer)
   145      method = method.id2name if method.kind_of?(Integer)
   146      ali = ali.id2name if ali.kind_of?(Integer)
   147
   148      self.send(:define_method, ali) do |*args,█|
=> 149        begin
   150          instance_variable_get(accessor).__send__(method, *args,█)
   151        rescue Exception
   152          [email protected]_if{|s| /^\\(__FORWARDABLE__\\):/ =~ s} unless Forwardable2::debug
   153          Kernel::raise

(rdb:1) step
/usr/lib/ruby/1.8/forwardable2.rb:150
instance_variable_get(accessor).__send__(method, *args,█)

Allready much, much better.

Of course, with the string-eval approach to metaprogramming Ruby itself could do better by saving the string that is being evaled to be able to refer to it later at step-through time. However currently we don’t have this option.

Tomáš Pospíšek

extending RedCloth markup

There are various approaches when trying to extend the Textile markup that RedCloth understands with own tags or syntax. Some approaches documented on the net have changed or don’t work any more, since RedCloth has been rewritten in Version 4.

Below is a fairly robust aproach, that is based on the assumption, that RedCloth leaves HTML tags inside the markup untouched and passes them on to the application consuming the translated markup.

The below code has been used in the Madek project. Madek, a Ruby On Rails application, uses irwi which provides a Wiki to Madek. And finally irwi uses RedCloth to do the rendering from Textile to HTML. That’s where we hook in.

We want to have a few special tags, which make the life of the Madek Wiki admin simpler, by letting him write the following inside the Textile markup:

[media=210      | Das Huhn]
[screenshot=210 | Das Huhn]
[video=210      | Das Huhn]

That will finally produce this HTML output:

<a href="/media_entries/210">Das Huhn</a>
<img src="/media_entries/210/image" title="Das Huhn"/>
<video src="/media_entries/210/image" title="Das Huhn"/>
  <a href='/media_entries/210'>(see video)</a>
</video>

Here’s the implementation:

class RedClothMadek

  ActionView::Base.sanitized_allowed_tags << 'video'

  def initialize
    require 'redcloth'
  end

  def format( text )
    ::RedCloth.new( replace_madek_tags(text) ).to_html
  end

  # Transforms the follwing Textile markups:
  # 
  #   [media=210      | Das Huhn] -> <a href="/media_entries/210">Das Huhn</a>
  #   [screenshot=210 | Das Huhn] -> <img src="/media_entries/210/image" title="Das Huhn"/>
  #   [video=210      | Das Huhn] -> <video src="/media_entries/210/image" title="Das Huhn"/>
  #                                    <a href='/media_entries/210'>(see video)</a>
  #                                  </video>
  #
  def replace_madek_tags( text )

    # unfortunately having multiple matches in gsub doesn't seem to work, therefore
    # we fall back to $1 $2
    #
    text.gsub(/[\s*media\s*=\s*(\d+)\s*\|\s*([^]]+)\s*]/) { |number,txt|

           "<a href='/media_entries/#{$1}'>#{h($2)}</a>"                 }.

         gsub(/[\s*screenshot\s*=\s*(\d+)\s*\|\s*([^]]+)\s*]/) { |number,title|

           "<img src='/media_entries/#{$1}/image' title='#{h($2)}'/>"    }.

         gsub(/[\s*video\s*=\s*(\d+)\s*\|\s*([^]]+)\s*]/) { |number,title|
           "<video src='/media_entries/#{$1}/image' title='#{h($2)}'>" +
             "<a href='/media_entries/#{$1}'>(see Wideo)</a>" +
           "</video>"  }
  end

end

And finally, irwi needs to be told to use that formatter instead of RedCloth directly. From config/environment.rb:

require "#{Rails.root}/lib/red_cloth_madek.rb"
Irwi.config.formatter = RedClothMadek.new

Tomáš Pospíšek

extending RedCloth markup

There are various approaches when trying to extend the Textile markup that RedCloth understands with own tags or syntax. Some approaches documented on the net have changed or don’t work any more, since RedCloth has been rewritten in Version 4.

Below is a fairly robust aproach, that is based on the assumption, that RedCloth leaves HTML tags inside the markup untouched and passes them on to the application consuming the translated markup.

The below code has been used in the Madek project. Madek, a Ruby On Rails application, uses irwi which provides a Wiki to Madek. And finally irwi uses RedCloth to do the rendering from Textile to HTML. That’s where we hook in.

We want to have a few special tags, which make the life of the Madek Wiki admin simpler, by letting him write the following inside the Textile markup:

[media=210      | Das Huhn]
[screenshot=210 | Das Huhn]
[video=210      | Das Huhn]

That will finally produce this HTML output:

<a href="/media_entries/210">Das Huhn</a>
<img src="/media_entries/210/image" title="Das Huhn"/>
<video src="/media_entries/210/image" title="Das Huhn"/>
  <a href='/media_entries/210'>(see video)</a>
</video>

Here’s the implementation:

class RedClothMadek

  ActionView::Base.sanitized_allowed_tags << 'video'

  def initialize
    require 'redcloth'
  end

  def format( text )
    ::RedCloth.new( replace_madek_tags(text) ).to_html
  end

  # Transforms the follwing Textile markups:
  # 
  #   [media=210      | Das Huhn] -> <a href="/media_entries/210">Das Huhn</a>
  #   [screenshot=210 | Das Huhn] -> <img src="/media_entries/210/image" title="Das Huhn"/>
  #   [video=210      | Das Huhn] -> <video src="/media_entries/210/image" title="Das Huhn"/>
  #                                    <a href='/media_entries/210'>(see video)</a>
  #                                  </video>
  #
  def replace_madek_tags( text )

    # unfortunately having multiple matches in gsub doesn't seem to work, therefore
    # we fall back to $1 $2
    #
    text.gsub(/[\s*media\s*=\s*(\d+)\s*\|\s*([^]]+)\s*]/) { |number,txt|

           "<a href='/media_entries/#{$1}'>#{h($2)}</a>"                 }.

         gsub(/[\s*screenshot\s*=\s*(\d+)\s*\|\s*([^]]+)\s*]/) { |number,title|

           "<img src='/media_entries/#{$1}/image' title='#{h($2)}'/>"    }.

         gsub(/[\s*video\s*=\s*(\d+)\s*\|\s*([^]]+)\s*]/) { |number,title|
           "<video src='/media_entries/#{$1}/image' title='#{h($2)}'>" +
             "<a href='/media_entries/#{$1}'>(see Wideo)</a>" +
           "</video>"  }
  end

end

And finally, irwi needs to be told to use that formatter instead of RedCloth directly. From config/environment.rb:

require "#{Rails.root}/lib/red_cloth_madek.rb"
Irwi.config.formatter = RedClothMadek.new

Tomáš Pospíšek

extending RedCloth markup

There are various approaches when trying to extend the Textile markup that RedCloth understands with own tags or syntax. Some approaches documented on the net have changed or don’t work any more, since RedCloth has been rewritten in Version 4.

Below is a fairly robust aproach, that is based on the assumption, that RedCloth leaves HTML tags inside the markup untouched and passes them on to the application consuming the translated markup.

The below code has been used in the Madek project. Madek, a Ruby On Rails application, uses irwi which provides a Wiki to Madek. And finally irwi uses RedCloth to do the rendering from Textile to HTML. That’s where we hook in.

We want to have a few special tags, which make the life of the Madek Wiki admin simpler, by letting him write the following inside the Textile markup:

[media=210      | Das Huhn]
[screenshot=210 | Das Huhn]
[video=210      | Das Huhn]

That will finally produce this HTML output:

<a href="/media_entries/210">Das Huhn</a>
<img src="/media_entries/210/image" title="Das Huhn"/>
<video src="/media_entries/210/image" title="Das Huhn"/>
  <a href='/media_entries/210'>(see video)</a>
</video>

Here’s the implementation:

class RedClothMadek

  ActionView::Base.sanitized_allowed_tags << 'video'

  def initialize
    require 'redcloth'
  end

  def format( text )
    ::RedCloth.new( replace_madek_tags(text) ).to_html
  end

  # Transforms the follwing Textile markups:
  # 
  #   [media=210      | Das Huhn] -> <a href="/media_entries/210">Das Huhn</a>
  #   [screenshot=210 | Das Huhn] -> <img src="/media_entries/210/image" title="Das Huhn"/>
  #   [video=210      | Das Huhn] -> <video src="/media_entries/210/image" title="Das Huhn"/>
  #                                    <a href='/media_entries/210'>(see video)</a>
  #                                  </video>
  #
  def replace_madek_tags( text )

    # unfortunately having multiple matches in gsub doesn't seem to work, therefore
    # we fall back to $1 $2
    #
    text.gsub(/[\s*media\s*=\s*(\d+)\s*\|\s*([^]]+)\s*]/) { |number,txt|

           "<a href='/media_entries/#{$1}'>#{h($2)}</a>"                 }.

         gsub(/[\s*screenshot\s*=\s*(\d+)\s*\|\s*([^]]+)\s*]/) { |number,title|

           "<img src='/media_entries/#{$1}/image' title='#{h($2)}'/>"    }.

         gsub(/[\s*video\s*=\s*(\d+)\s*\|\s*([^]]+)\s*]/) { |number,title|
           "<video src='/media_entries/#{$1}/image' title='#{h($2)}'>" +
             "<a href='/media_entries/#{$1}'>(see Wideo)</a>" +
           "</video>"  }
  end

end

And finally, irwi needs to be told to use that formatter instead of RedCloth directly. From config/environment.rb:

require "#{Rails.root}/lib/red_cloth_madek.rb"
Irwi.config.formatter = RedClothMadek.new

Tomáš Pospíšek

annotating third party web pages

Problem: I have a web site/page that I visit regularily which I want to annotate with my notes.

More specifically, I was regularly searching through the Homegate real estate hub looking for a new home. It goes without saying that I was again and again forgetting which objects I had already looked at, which objects were really interesting and I should check out more closely.

Therefore the need to annotate search results.

The here presented approach should be applicable for annotation of other web pages as well. It’s based on the observation, that restful web applications need to operate with asset IDs. These asset IDs can be reused to enrich the asset locally with additional data, such as notes. Thus we’re looking for those IDs in specific elements of the page and add a bit of HTML markup to those places.

Of course that aproach only works as long as the web site doesn’t heavily change its markup and doesn’t rendomly change asset IDs.

Here’s a screenshot of regular Homegate search results:

And here’s the same page after scripting it with Greasemonkey:

You’ll notice the input field with the comment in it.

The idea is simple: add an input box to each search result, where you write your comment. When the focus leaves the input box, the comment is stored to localStorage.

Starting with Greasemonkey is quite easy. There are howtos and templates to start from, such as this one.

However, regular JavaScripting and Greasemonkey JavaScripting do not work in exactly the same way:

One of the differences is that Greasemonkey creates a separate JavaScript environment, in which the Greasemonkey scripts are executed. That is calling Javascript contained in the page from Greasemonkey and the inverse calling Greasemonkey scripts from the page is not possible by default. This is on purpose, so that the web page can not detect and not interfere with the Greasemonkey scripts, because you want your Greasemonkey scripts to work allways on some page, whether or not that page likes it or not. Therefore no interference is possible.

Greasemonkey however provides a standard way to access the web page’s scripts, and that’s through the “unsafeWindow” object, which is a reference to the web page’s environment.

I had two mechanisms I had to make accessible using the “unsafeWindow” handle:

  • the first was accessing JQuery, which is included by default by the Homegate page. Since I needed to use JQuery functions in my Greasemonkey script, I got a reference to it via the standard Greasemonkey precedure:
    var jQuery = unsafeWindow['jQuery'];
  • the second mechanism that needed to cross the boundaries between the web page and Greasemonkey was callbacks from the web page to my Greasemonkey script. This is necessary, because I’m attaching “input” elements to each search result, which contain a note and which, “onblur”, need to call a function that saves the content of the input box. Here’s part that constructs the input element:
    jQuery("<input onclick='event.cancelBubble = true;'" +
                 " onblur='saveComment(this, immoID);'>").insertAfter(immoElement);

And this is the function that gets called back by “onblur”:

    unsafeWindow.saveComment = function(element, immoID) {
      unsafeWindow.localStorage.setItem(immoID, element.value);
    };

The next interesting thing you’ll note is usage of ‘locaStorage’. Support for the latter in browsers does not seem mature yet. One problem I’ve encountered when developing under Firefox 3.6 was that saving to ‘localStorage’ was not possible when cookies were disabled (see this report). Thus you’ll need to permanently enable cookies for Homegate in order for the script to be able to save its data.

Finally, while developing, a major problem was, that Firefox did not show me errors in the Greasemonkey scripts. Thus either the script would work or not work and fail completely silently. That made debugging a bit painful.

So now, here’s the script.

Tomáš Pospíšek

PS: This script also lives at userscripts.org

annotating third party web pages

Problem: I have a web site/page that I visit regularily which I want to annotate with my notes.

More specifically, I was regularly searching through the Homegate real estate hub looking for a new home. It goes without saying that I was again and again forgetting which objects I had already looked at, which objects were really interesting and I should check out more closely.

Therefore the need to annotate search results.

The here presented approach should be applicable for annotation of other web pages as well. It’s based on the observation, that restful web applications need to operate with asset IDs. These asset IDs can be reused to enrich the asset locally with additional data, such as notes. Thus we’re looking for those IDs in specific elements of the page and add a bit of HTML markup to those places.

Of course that aproach only works as long as the web site doesn’t heavily change its markup and doesn’t rendomly change asset IDs.

Here’s a screenshot of regular Homegate search results:

And here’s the same page after scripting it with Greasemonkey:

You’ll notice the input field with the comment in it.

The idea is simple: add an input box to each search result, where you write your comment. When the focus leaves the input box, the comment is stored to localStorage.

Starting with Greasemonkey is quite easy. There are howtos and templates to start from, such as this one.

However, regular JavaScripting and Greasemonkey JavaScripting do not work in exactly the same way:

One of the differences is that Greasemonkey creates a separate JavaScript environment, in which the Greasemonkey scripts are executed. That is calling Javascript contained in the page from Greasemonkey and the inverse calling Greasemonkey scripts from the page is not possible by default. This is on purpose, so that the web page can not detect and not interfere with the Greasemonkey scripts, because you want your Greasemonkey scripts to work allways on some page, whether or not that page likes it or not. Therefore no interference is possible.

Greasemonkey however provides a standard way to access the web page’s scripts, and that’s through the “unsafeWindow” object, which is a reference to the web page’s environment.

I had two mechanisms I had to make accessible using the “unsafeWindow” handle:

  • the first was accessing JQuery, which is included by default by the Homegate page. Since I needed to use JQuery functions in my Greasemonkey script, I got a reference to it via the standard Greasemonkey precedure:
    var jQuery = unsafeWindow['jQuery'];
  • the second mechanism that needed to cross the boundaries between the web page and Greasemonkey was callbacks from the web page to my Greasemonkey script. This is necessary, because I’m attaching “input” elements to each search result, which contain a note and which, “onblur”, need to call a function that saves the content of the input box. Here’s part that constructs the input element:
    jQuery("<input onclick='event.cancelBubble = true;'" +
                 " onblur='saveComment(this, immoID);'>").insertAfter(immoElement);

And this is the function that gets called back by “onblur”:

    unsafeWindow.saveComment = function(element, immoID) {
      unsafeWindow.localStorage.setItem(immoID, element.value);
    };

The next interesting thing you’ll note is usage of ‘locaStorage’. Support for the latter in browsers does not seem mature yet. One problem I’ve encountered when developing under Firefox 3.6 was that saving to ‘localStorage’ was not possible when cookies were disabled (see this report). Thus you’ll need to permanently enable cookies for Homegate in order for the script to be able to save its data.

Finally, while developing, a major problem was, that Firefox did not show me errors in the Greasemonkey scripts. Thus either the script would work or not work and fail completely silently. That made debugging a bit painful.

So now, here’s the script.

Tomáš Pospíšek

PS: This script also lives at userscripts.org

annotating third party web pages

Problem: I have a web site/page that I visit regularily which I want to annotate with my notes.

More specifically, I was regularly searching through the Homegate real estate hub looking for a new home. It goes without saying that I was again and again forgetting which objects I had already looked at, which objects were really interesting and I should check out more closely.

Therefore the need to annotate search results.

The here presented approach should be applicable for annotation of other web pages as well. It’s based on the observation, that restful web applications need to operate with asset IDs. These asset IDs can be reused to enrich the asset locally with additional data, such as notes. Thus we’re looking for those IDs in specific elements of the page and add a bit of HTML markup to those places.

Of course that aproach only works as long as the web site doesn’t heavily change its markup and doesn’t rendomly change asset IDs.

Here’s a screenshot of regular Homegate search results:

And here’s the same page after scripting it with Greasemonkey:

You’ll notice the input field with the comment in it.

The idea is simple: add an input box to each search result, where you write your comment. When the focus leaves the input box, the comment is stored to localStorage.

Starting with Greasemonkey is quite easy. There are howtos and templates to start from, such as this one.

However, regular JavaScripting and Greasemonkey JavaScripting do not work in exactly the same way:

One of the differences is that Greasemonkey creates a separate JavaScript environment, in which the Greasemonkey scripts are executed. That is calling Javascript contained in the page from Greasemonkey and the inverse calling Greasemonkey scripts from the page is not possible by default. This is on purpose, so that the web page can not detect and not interfere with the Greasemonkey scripts, because you want your Greasemonkey scripts to work allways on some page, whether or not that page likes it or not. Therefore no interference is possible.

Greasemonkey however provides a standard way to access the web page’s scripts, and that’s through the “unsafeWindow” object, which is a reference to the web page’s environment.

I had two mechanisms I had to make accessible using the “unsafeWindow” handle:

  • the first was accessing JQuery, which is included by default by the Homegate page. Since I needed to use JQuery functions in my Greasemonkey script, I got a reference to it via the standard Greasemonkey precedure:
    var jQuery = unsafeWindow['jQuery'];
  • the second mechanism that needed to cross the boundaries between the web page and Greasemonkey was callbacks from the web page to my Greasemonkey script. This is necessary, because I’m attaching “input” elements to each search result, which contain a note and which, “onblur”, need to call a function that saves the content of the input box. Here’s part that constructs the input element:
    jQuery("<input onclick='event.cancelBubble = true;'" +
                 " onblur='saveComment(this, immoID);'>").insertAfter(immoElement);

And this is the function that gets called back by “onblur”:

    unsafeWindow.saveComment = function(element, immoID) {
      unsafeWindow.localStorage.setItem(immoID, element.value);
    };

The next interesting thing you’ll note is usage of ‘locaStorage’. Support for the latter in browsers does not seem mature yet. One problem I’ve encountered when developing under Firefox 3.6 was that saving to ‘localStorage’ was not possible when cookies were disabled (see this report). Thus you’ll need to permanently enable cookies for Homegate in order for the script to be able to save its data.

Finally, while developing, a major problem was, that Firefox did not show me errors in the Greasemonkey scripts. Thus either the script would work or not work and fail completely silently. That made debugging a bit painful.

So now, here’s the script.

Tomáš Pospíšek

PS: This script also lives at userscripts.org

JMeter Series

This is a short series of howtos for and a critique of JMeter v2.4.

The following articles have been done:

Tomáš Pospíšek, 4.1.2011

Waiting for a page change in JMeter

(This article is part of the JMeter Series)

While testing a Rails application there was a situation where a background worker (DJB or background task) would get an order to execute and eventually the completed order would appear on a page.

Thus we needed to wait for the page to be updated and continue the test after.

User Defined Variables

I’m predefining used variables here - see “We set up our variables” in the Extracting text from a page and using it somewhere else in JMeter article for an exaplanation why.

Clear Loop Variable

This step is not strictly necessary in our example here, however if you want to reuse this sequence of steps more than one time - that is call it from different places, then you need to clear the loop variable first.

Debug

Since creating these steps was not easy, I’ve put a debug statement in so that I see, when JMeter calls into these steps and to see how the variables are set.

The While Controler

Here we really start the loop.

I’m not sure whether if would be better to use BeanShell commands here instead - using JavaScript to test the loop condition works in any case…

Getting the page

We get the page we want to check.

Extract the data we want to check from the page and assign it to our loop variable

Make sure our request is not being cached

It’s better - as far as we can - to be sure we force the web application to recreate the page we’re waiting for to change. I’m not sure I’ve done it right here, however it works:

Wait a bit before re-looping

We wait two seconds here before retrying.

You can download this JMeter test from here

Tomáš Pospíšek, 4.1.2011

Extracting text from a page and using it somewhere else in JMeter

(This article is part of the JMeter Series)

In the following we’ll do these things:

  1. we go to a form submit page
  2. we order something
  3. we get that something we ordered

In more detail:

We use a “Cookie Manager” to carry forward cookies between calls

We set up our variables

The idea here being that we predefine the variables, so they would show up in the “Debug Sampler” which makes debugging easier, because you see at each step, whether the variable has the correct value or not. I do predefine the variables with a “non-value” so that I can see immediately whether the variable has been already assigned something during the execution of the test or not.

We go to a form submit page

We extract the AUTHTOKEN from that form

(The auth token is being used by the web application to prevent cross-site-scripting)

Notice that the extractor is set to extract the string from the HTTP reply-body.

We also extract the session cookie

This is because we want to pass the cookie to an outside application, so that it can call the web app from within the same session.

Notice that this time we told the extractor to extract the string from the HTTP reply-headers.

We now use all our extracted parameters to submit our order

Note that:

  • we are passing the AUTHTOKEN along with the HTTP POST
  • since we are not sure whether possibly the AUTHTOKEN contains some problematic character (in my case it was an equal sign ‘=’ that was interferring with the parameter encoding) we tell JMeter to URL-encode the string.
  • we also pass along another parameter telling the web app what we’re interested in getting back from our order
  • we are using HTTPS

Submitting the order will redirect us to the page that’ll show us the resulting order.

We extract the ORDER_ID from HTTP headers which would otherwise redirect us to the resulting page

The headers contain the location of the resulting order page where we would be redirected to.

Finally we download the resulting artefact with curl

Note that:

  • we are passing the SESSION_COOKIE to curl to be able to download the artifact in the same (potentially authenticated) session.

  • we also construct the download URL from the ORDER_ID

You can download this JMeter test from here

Tomáš Pospíšek, 4.1.2011

The JMeter "Workbench", a trapdoor for the newbie

(This article is part of the JMeter Series)

Upon starting JMeter you’ll see two branches: “Test Plan” and “WorkBench”.

“Test Plan” is the place where your tests will live.

What the purpose of “WorkBench” is, is not really clear. It seems to be meant to be a place to do your throw-away experimentation.

The really crucial “trap” of the “WorkBench” is however, that JMeter will throw away whatever you put into the “WorkBench” upon exit. JMeter will not save the contents of the “WorkBench” if you tell it to “Save” your work and you’ll loose whatever is in there. It will only save the contents of the “Test Plan” branch.

So be ware of putting anything in the “Work Bench”. You’re bound to get burned.

Tomáš Pospíšek, 27.12.2010

Making your JMeter Test modular

(This article is part of the JMeter Series)

As tests get larger, or as steps need to be repeated you’ll want to structure your tests into distinct entities - these seem to be called “Modules” in JMeter.

However, there is no “Module” element JMeter. As a “Module” you can however use the “Simple Controller”. It allows you to drop other elements into it and to name them as a whole. I don’t know whether it has additional features such as providing scoping of any kind.

Thus if you need to “call” the same set of steps from different places, then you can place them in a “Simple Controller”. However you’ll need to place that controller somewhere and as such it will get executed in that place and order. You can prevent it being executed by disabling it:

even though being disabled, components inside a Controller still can be called and executed. Therefore, you can use a disabled “Simple Controller” as a repository for reusable components or modules:

On the picture you can see a “module repository” and a call to a specific “Module” inside it.

Tomáš Pospíšek, 30.12.2010

Debugging JMeter Tests

(This article is part of the JMeter Series)

Useful ways to debug JMeter as far as I know:

  • insert a “Debug Sampler”

the “Debug Sampler” will emit “everything that’s known” to JMeter. That output can be displayed in the “View Results Tree Listener”:

  • have a look at the JMeter log file (which is usually dumped from where you’ve started JMeter)

  • check the output of one of the listeners such as the “View Results Tree Listener”.

  • if you need to debug regexes of a “Regular Expression Extractor” then you can have a look at the source code of the page you want to extract a value from inside the “View Results Tree Llistener” and enter the same regular expression in the “Search” field (searching is only implemented in JMeter > v2.4).

In that picture you can see JMeter matching and displaying the regex entered in the search field inside an HTML page retrieved though the “HTTP Client Sampler”.

Tomáš Pospíšek, 1.1.2011

Some words on overall usefulnes of JMeter

(This article is part of the JMeter Series)

Purpose of JMeter

JMeter is a testing tool. It comes accross as a graphical tool, where you can half visually half through text define your test. Its roots seem to lie in web testing - that means testing a website on how long it takes to return pages, how well it does under stress, how well it scales with increasing numbers of parallel requests etc.

However JMeter can be both used non-visually and has extended far beyond web testing.

Stability of JMeter

While working on a test with JMeter 2.3.4 it has had a multitude of stability problems:

  • running out of heap space and crashing
  • running out of heap space while testing and thus not being able to correctly execute and finish tests
  • hanging and not being responsive any more
  • not being able to terminate its internal threads
  • not terminating launched outside processes
  • while saving the work damaging the saved file itself and rendering it unreadable and thus loosing the work

That means that you can expect JMeter to hang or crash at any point while developing, executing or saving your tests.

That means that you should really be saving (CTRL-S or Apple-S) very often. On the other hand saving often will increase the likeness of your save file itself being destroyed and your work lost (see last point above). Thus saving is not enough - you need to version your work too, in order to be able to access older versions, that are not damaged.

Running something like this from a Unix command line should be saving revisions of your current work file - you’ll need to be saving your work continuously though:

  $ revision=99
  $ save_interval=300 # seconds
  $ work_file="Load Test.jmx"
  $ while true; do
  >    cp "$work_file" "$work_file.$revision"
  >    echo "Saved "$work_file" under "$work_file.$revision"
  >    sleep $save_interval
  >    if [ "$?" != 0 ]; then break; fi
  > done

One problem that does not seem to be resolved within JMeter at all is “big” requests. The various Samplers seem allways to put whatever the get from the test target into memory. Thus doing tests on multimedia content with gigabyte sized files will make JMeter run out of memory. There does not seem to be a way to tell JMeter to throw away downloaded content and/or to treat it in flight and not to save it in its entirety.

Documentation

The online JMeter documentation is brief and is rather just mentioning features than describing them in depth. Also, it’s not easy at all to find ready made examples that demonstrate syntax and finer points of how to use JMeter. When accessing the help coming with JMeter itself the JMeter instance would just hang.

I did a lot of searching the web, without much success and a lot of trying.

Debugging

What do you do when tests don’t run the way you want or do unexpected things? Then you need a way to debug them. There are three debugging facilities of JMeter, that are not hard to access:

  • the Debug Sampler
  • the JMeter log file
  • the various Listeners

Here’s a snapshot of the Debug Sampler:

And a snapshot of the Debug Sampler displaying JMeter properties inside the “View Results Tree” Listener.

These facilities are very useful, however the various JMeter elements themselves are blackboxes - there seems to be no way to introspect them or “step through” them at runtime. When they “don’t work”, then either they might display something useful inside the JMeter Log or possibly you’ll be able to collect hints of why they might not work after they’ve run through the Debug Sampler results. If both approaches fail, then it’s not clear what else can be done other than pluging into the source code of JMeter itself.

Complexity

I’d describe working JMeter as “to do easy things is non-trivial and doing hard things is extremely demanding”

The various testing elements, such as “Logic Controllers”, “Listeners” etc. seem to have different scopes and different orders of execution.

“Listeners” for example seem to be global in scope, in that they “see” and can react on whatever happens in the whole of JMeter. They might be possibly limited to a “Thread Group” though, depending on where they are placed - I did not try to find out.

Doing slightly more complex things - such as a loop to repeat a row of steps multiple times, or executing external scripts does not seem to be supported within JMeter itself - one needs to ressort to nearby tools such as the Bean Shell and accordingly to learn the syntax and the working of that tool as well.

Same goes for “Extractors” which refer to a nearby tools that do Perl regexes or XPath matching.

In short to do even rather easy tests one needs to get to know a vast array of tools and syntaxes and how those various tools interact with each other.

The pros

All those negative points seem to weight quite heavily and they do. However after quite a lengthy and steep learning curve JMeter as a tool starts showing its strength. Once one has resolved a larger number of problems, adding more steps and functionality to JMeter starts getting easier and one productivity starts to increase.

It’s also fun working with such a complex, versatile and powerful tool and it’s fun to work with a tool that starts from a fresh and different perspective how to solve (visually) problems.

Conclusion

One thing that one gets for “free” with JMeter is the nice reports generated by the various included Listeners. It’s something that will certainly impress management and is useful for gaining a high level understanding of the performance of a web-site.

Other things - like traversing web sites, extracting and matching information might be just as well done with specialised tools coming along with Python, Ruby or command line tools available under Unix.

Ease of automation when using proper scripting languages and their effectiveness and efficiency will be hard to beat given that one allready has a good working knowledge of them.

To me it’s not really clear whether using a scripting language for the task of testing and profiling is not more efficient and more flexible than using JMeter.

However I do reccomend JMeter: it’s fun to work with and it does give a new perspective on how things can be done.

Tomáš Pospíšek, 28.12.2010

Back to Top

Sustaining Members