Just an example of the assembly structure I wanted to have:

• Main.exe
  • lib (folder)
    • Shared.dll (used by Main.exe, Plugin1.dll and Plugin2.dll)
    • Plugin1.dll
    • Plugin2.dll

The answer – open up your app.config from Main.exe project and add the following (under <configuration> tag):

<runtime>
  <assemblyBinding xmlns="urn:schemas-microsoft-com:asm.v1">
    <probing privatePath="lib"/>
  </assemblyBinding>
</runtime>

That is it. Now it will also search for referenced libraries in the lib folder

ClearType vs. GDI++

Well, I think that the difference is pretty obvious. So if you are ready to start you new font life - go here and download the file named gdi0870.zip (404kb). Extract anywhere you want and launch gditray.exe. Choose 'enable' from the context menu of the tray icon and voila - your font are rendered without the help of that ugly, ugly ClearType.

So, this is the way you make sessions/logins work in ruby:

class SessionController < ApplicationController
  def login
    @session['logged'] = true
  end
  def logout
    @session['logged'] = false
  end
end

The @session variable is available anywhere. Use it anywhere. For example you could come up with the following view showing whether the user is logged in:

<p>Logged in: <%= @session['logged'] %> </p>

And that is it. Anything beyond that (like adding a User model etc.) is out of this post's target. There are many articles on those things yet no clear explanation where to find and how to init a session. As you can see - the answer is clear - you don't The session is already there. Just start using it.

1. You create a new HTML file in komodo and save it in some directory.
2. You double-click the script in the Komodo Edit toolbox.
3. The script adds the following folders/files to the project (paths relative to the file currently in editor!)
   • css/
   • css/style.css
   • css/reset.css
4. Also the file adds the following HTML code at the cursor in current opened file:

Voila. The needed action is done.

Now, the snippet code:

import sys
import shutil
import os
import os.path

# argv[1] - dir of cur project (%D)
# argv[2] - current file name (%F)
if os.path.exists(sys.argv[1]+'/css/') != True:
    os.mkdir(sys.argv[1]+'/css/')

# NB! Edit the paths to style.css and reset.css
# (these will be copied into your project)!
shutil.copy('C:/Work/__distributives/css/reset.css',sys.argv[1]+'/css/')
shutil.copy('C:/Work/__distributives/css/style.css',sys.argv[1]+'/css/')

print '
'
print '
'

Save this as a file somewhere (e.g. C:/Work/__scripts/add_style_css.py).
Open up Komodo Edit and add it to your toolbox (Toolbox->Add->New command...) with the following settings:

• Command: python add_style_css.py %D %F
• Insert output: check
• Start in: whatever the script resides in (e.g. C:/Work/__scripts)
• anything else should bet left as it is

Enjoy

These are the features we will try to get into our menu:

1. It will be horizontal
2. It will have splitters (made with images)
3. It will have active class for rollover / selected item (background change)
4. Menu items will be of different size

This code works for Internet Explorer 6+, Firefox 2.0+, Opera 9+, Safari (at least on Safari 4 - tested on Windows version of it).

Here is a mockup that we have:

Menu mockup

NB! If you just want to see how to do it - go here for the demo / full code.

Ok, first, the HTML you need for this thing to happen:

<ul id="menu">
 <li><a href="#">Menu item</a></li>
 <li class="active"><a href="#">Longer item</a></li>
 <li><a href="#">Short</a></li>
 <li><a href="#">Finally</a></li>
</ul>

This is the needed thing. One more thing to add - if you lack one element for background (if <li> and <a> are already having backgrounds and you need another one) - just wrap <a> with a <span>.

The CSS code for the menu:

html, body {
    font-family: "Trebuchet MS";
    font-size: 16px;
}
a       { text-decoration: none; color: #ffffff; }
a:hover { text-decoration: underline; }

#menu {
    margin:0;
    padding:0 12px;
    list-style-type: none;

    background: url(menu-bg.png) no-repeat;
    height:33px;
}
#menu li {
    float: left;
    display: block;
    margin-top:1px;
    background: url(menu-item-bg.png) no-repeat top right;
}
#menu li a {
    display: block;
    padding:4px 12px 4px 10px;
}
#menu li.active {
    background: url(menu-item-active-bg.png) no-repeat;
}
#menu li.active a {
    background:
        url(menu-item-active-bg-right.png)
        no-repeat top right;
}

Some explanations: paddings and heights are just to make this thing look better a bit. The most important things are float, display and background aligns! If you want to create the menu yourself - just change the images and change the paddings / heights to correspond.

Now, the list images we will have to use:

Menu background

Menu item background

Active menu item background

Right-handed chunk

Of course you may want it to be a rollover effect. This is very easy to accomplish by using JavaScript. I will use jQuery for this but you can pretty much use any framework or just plain JavaScript to do that. The code for the rollover effect in jQuery is as follows:

    $(document).ready(function () {
       $('#menu li').mouseenter(function ()
           { $(this).addClass('active'); });
       $('#menu li').mouseleave(function ()
           { $(this).removeClass('active'); });
    });

Finally, the whole stuff compiled together

On the other hand I must say that this craze is a bit over the edge. Why is that and how can it be seen? Well, from a dev perspective this mostly conserns HTML. Everybody wants valid (X)HTML today and I am not an exception. The other side of this is that everyone wants <div> vased websites with <ul> menus. Full stop. Now why is that? I admit, yes - <table> based sites are horrible to look at (the source code I mean). Yes, it is much clearer to do sites using <div> elements. But can anybody explain to me what is this "oh-so-obvious" reason for making menus using
<ul><li> menus? I don't mind it, but let us be honest - this can be a pain sometimes when you want a decent menu.

I see a lot of enthusiasts voting for <ul> who actually just repeat after others. They always use very simple menu examples to indicate how good this costruct is and I do not blame them. But is there any reason to hate <table> in menus? I do not see it. Do you? Using tables for menus is even a bit more natural to me than using lists - they are much more versatile in certain situation and demand much less effort to make them look good. Lists are an idealistic view at the source code structure and are good, but when the customer requires a more complex menu (not just plain background and color changing on hover) I often come to a conclusion that tables are the way to go.

I am, of course, not saying that tables should be used in every single menu, but I want to point is that tables are nothing to be afraid of. Do not listen to the masses, who scream one phrase all the time - think for yourself and decide which concept to use in a certain case of yours. Be reasonable.

Remember how you used to write those small pure HTML pages without caring about whether it passes the validity tests? Remeber how you didn't care about Mozilla, Opera and Safari? You only had to make sure that this will work in IE and put a small text on the front page - "This site was made for IE6".

After that came PHP, which was overwhelming. It is a great tool designed to make scripting pages with ease. Then came databases and object-oriented paradigms. That is the time when I first started to feel frustrated. The code did not fit into my head anymore. I had to review the classes I have, the functions that they have, members, deal with connections and other crap.

And the latest thing - AJAX. This thing has been terrorising every single web developer. When I hear AJAX I always have two conflicting feelings - it is cool and it is very unpleasant to make it happen. I mean JavaScript was so simple (and mostly unnecessary) until AJAX hit the internet (and Web 2.0, of course). All you did was getElementById and validate forms with it sometimes.

Where am I going with this? Well... It is simple. At the moment the whole web development process with HTML/PHP/JS is quite complicated if you want to make something exceptional. As I work I keep seeing that I am performing repeating actions now and then. And I think that now it is time to make something new. A new way, a new tool, tailored for the specific needs of the web developer community. I will be not surprised if it will be something ASP.NET-like. Let it be. It is time to either dramatically improve the web development tools we have now or to create a new all-in-one tool that takes the concepts of HTML, PHP and JS/AJAX together and puts them into a universal form.

Tasks

• slicing designs into images and cobining it to (X)HTML/CSS
• script writing (PHP / JavaScript)
• debugging
• testing IE compatibility
• quick editing
• database handling
• OS: Microsoft Windows
• others

Slicing

Adobe Photoshop - using it mostly because the designs are in PSD files. I do not actually slice anything here. I just need this to take screenshots of the design to slice it with Paint.NET afterwards.

Paint.NET - a free, but absolutely awesome graphical editor without the tremendous amount of unnecessary functionality. I paste the screenshots from Photoshop here and select-and-crop them for slicing.

Creating (X)HTML/CSS

Adobe Dreamweaver - the most usable markup/style editor I have ever seen so far. Mostly because its autocomplete functions are so well thought-out. It is just a pleasure to use it. It is really worth the money Adobe wants people to pay for it.

Script Writing

Komodo Edit - a free tool, that has all the features I need. This is the only editor I know that has every feature that one might want in a script editor (PHP / JS autocomplete, code folding and project files) and does not have all those features that a regular person would use once a year max. Also, it is a very neat that Komodo has addons, which are implemented the same way as they are in Firefox. Very handy thing to have packed into a very familiar interface. Has the vertical selection feature, which is very important to me (just hold donw Alt+Shift while selecting). Can also be quite handy when need to change something in HTML/CSS and you do not want to bother opening up Dreamweaver.

Debugging

Firefox + Firebug and Web Developer addons - the best HTML/CSS/JS debugging tool. If you are not using those yet - you are now a web developer

I do not use any PHP debugging stuff. Still doing it the crappy way - echo'es and print_r's. And the messages from the PHP interpreter, of course.

Testing IE compatibility

Of course I test my sites in Firefox, Opera and Safari. But the only problem is Internet Explorer (I test 6, 7 and 8), so...

Multiple IE - adds shortcuts to IE versions (starting at version 3) to your quick-launch. Has PNG fix support. Nothing special, but I like it more than IETest.

Quick Editing

Notepad++ - hands down the most handy editor for some quick edits. Not for real production though, since there is just too much tweaking to be done in order to make it into a half-decent environment to work with. Editing some notes, viewing sources or creating a fast JS - Notepad++ makes it all happen for me.

Database Handling

MySQL GUI Tools - mostly the Query Browser and Administrator. No real alternatives here, so there is not much to argue about. Better than phpMyAdmin anyway

phpMyAdmin - the standart in hostings with MySQL databases, so I am simply forced to use it.

OS: Microsoft Windows

This is quite important to me. Even though I really lack multiple workspaces in Linux (and do not even dare telling me about those crappy emulators for Windows of this feature) I still think that Windows is a better enviroment for web developing. Actually, I think that some people go to do this on Linux just because there are tools on Linux that allow it (e.g. ADs cannot go there since there is no Photoshop there). It is just a trend. Want to be cool? Install Linux.

Well, I personally think that this is not true. Web development is much more comfortable for me in Windows because of the wide selection of tools (notice Notepad++, Dreamweaver and Paint.NET as well as the presence of normal IE). I do not dislike Linux though (I even have my lovely Ubuntu 9.04 installed). It is just that it does not provide me with all the tools to do my job in the most efficient way. Period.

Others

Of course there is a lot of other cool tools I use like Filezilla for FTP, Dia, MySQL Workbench and Visual Paradigm for modelling, Evernote for quick notes and so on. But they are quite small and sometimes completely unrelated to the topic of web development. Maybe someday I can write something about those little things that also help making the development process much more easy and less tiring.

Now, having some experience in trying to do database handling manually (e.g. having to use mysql_connect etc all the time) and letting Doctrine to do this for you I have one phrase to tell you: if you are not using ORM already you are losing a lot of precious time.

So, just as a small example of ORM godlyness. Really small and really simple. We need to create a page which holds orders from clients. Each order consists of items (duh!). This means two database entities - Orders and Items

First, let us see what we would need to do if we did not use Doctrine (or some another ORM):

• Create an SQL script for table creation (the main problem here is that SQL scripts are unreadable at all):
  • CREATE TABLE 'order' (id INT NOT NULL AUTO_INCREMENT, client_name VARCHAR(255), client_address VARCHAR(255));
  • CREATE TABLE 'item' (id INT NOT NULL AUTO_INCREMENT, title VARCHAR(255), cost FLOAT, order_id INT NOT NULL);
• Create classes Order and Item. This is always very annoying for one reason - you have to rewrite CRUD functions over and over and over and over again. I hate it. I am not going to write the CRUD support for those classes right now, but if you did it once - you do not want to do it again.

Well, that was pretty simple, but 1) time-consuming; 2) cofusing (you cannot see the overall scheme of things; 3) it is Boring.

Let's take a look how you can do this whole thing with Doctrine in no time. The basic idea is to create a YAML (*.yml) file, which describes your database scheme. Here is how our case would look like:

      Order:
        columns:
          client_name: string(255)
          client_address: string(255)
      Item:
        columns:
          title: string(255)
          cost: float
          order_id: integer
        relations:
          Order:
            local: order_id
            foreign: id
            foreignAlias: items

Voila. This is it. After that the only thing you have to do is type "php doctrine build-all-reload" in your favourite console and let Doctrine create the tables in the database and the classes for you. Right after that you can start using those to implement your business logic without caring about the CRUD functions!

In my script I could go like this right away:

      // Create an order and set its vars
      $order = new Order();
      $order->client_name = "Vasili Pupkin";
      $order->client_address = "Bobrujsk";
      // Add a new item
      $order[0]->title = "Saaremaa Vodka 80";
      $order[0]->cost = 15;
      // Add another item (which plays very well with the first item)
      $order[1]->title = "Good pickles";
      $order[1]->cost = 2;
      // Save the order to the database
      $order->save();

And that is the end of that story. Create a YAML file and start implementing the business-logic instead of spending a couple of hours (or more) writing boring monkey-oriented code.

This time I am going to go a bit deeper into minuses of MODx. Why? Well, all the pluses are available on the official website and I do not see the reason for repeating those. This post does not in anyway mean that MODx is a bad product! It just has some bad sides that I would find very useful to read about before starting to build a website based on this system. This post is healthy citicism and maybe a precaution for those wondering what can't you do with MODx.

1. Documentation

One of the most important minuses I have noted about MODx and anything related to documentation. Well, it is horrible. That concernes practically eveything. MODx heavily relies on customization by adding such things as chunks, snippets, plugins and modules. Not only the documentation of the MODx core system is horrible (and can be shown as agood example of how to not write documentation) but this practice also concerns anything starting from snippet usage to writing your own MODx components.

Now, I undertand that MODx did not even hit version 1.0 yet, but this (boring) part of any project is crucial to its success as the will to start using any framework-like product mostly depends on the features offered (50%) and documentation (50%). This is obvious and I do not understand why the MODx team is not concerned about this.

In the end I found myself googling such things as "wayfinder placeholders" and "ditto parameters" every now and then, since I keep forgetting those details. In addition to that all the documentation is done in different formats and designs, which is very annoying.

As a solution it would be not so hard to make some kind of a public documentation repository available to the MODx contributors with unified rules about documenting their code. Some kind of MSDN for MODx (not that complex of course).

Or as for now (e.g. until the 1.0 release) there should be at least a unified refence on all bundled MODx components!

2. Administrating area (manager) customization

At the moment the admin zone of MODx is pretty much monolithic. It cannot be changed fundamentally. And this is a major drawback, since admin zone is a vast part of the website itself. You cannot hide all of the stuff here and then, you cannot make it look too user-friendly after all (yes, I know about roles, groups etc. - it is far from ideal). Adding new custom functionality to the admin area can be done through modules, though the only place you can actually add this functionality is to the "Modules" menu (I get the logic, but I do not get why is there such a limit?).

The admin area must be fully customizable. Every single piece of it. Without editing the core MODx files.

* Many MODx people can, of course, say that there is MODx Revolution (which looks very promising), but at this point I will not discuss anything connected to it as the latest alpha was very buggy (right from the start) and unusable.

3. Templating engine

As already said in the previous post about MODx, the templating engine is quite far from being perfect. It has most of the necessary features, but far from Smarty by its functionality. I, personally, lack logical constucts ({if}). BUT! The behaviour of the templating engine can be changed. You can hook into the output event and modify the output as you wish (e.g. implementing as {if} would be quite easy, in fact, I have done it already as I cannot live without it and I do ot know how people use MODx without it - this simple plugin will be available as soon as I have time to make it look pretty).

Again, there is NO way you can mimic Smarty {if} with snippets - they simply work on a different level of MODx.

Conclusion

To conclude... Well, MODx is still pretty solid CMS system, which offeres a lot of features and works as a charm if configured and used corectly. Its disadvantage in the documentation area is, obviously, an effect of its open-source nature. The templating engine problems can be solved by creating a plugin (although it is quite strange to me that Smarty-like features are not present in MODx). Therefore, the only major problem at the moment is the manager customization. This problem is promised to be fixed when Revolution hits the ground but from what I've seen in the latest alpha it won't be a reliable product in near future. Therefore - modules are the only way to affect the admin area without going inside the core files of MODx.