And then to know — is it convenient to 2016-2018-2020 year will do it? Because PHP already does development not getting.
<?php $a = "2d9"; $a++; echo $a; → 2e0</property-->
<br / >
a — string. Increment line (which is not an entry in 10чной SCH) is <a href="http://php.net/manual/ru/language.operators.increment.php">on the principle of Perl</a>: "A"++ = "B", "Z"++ = "AA", "AA"++ = "AB". So we get a "2e0".<br>
<br / >
the <pre><code class="php"><?php $a = "2d9"; $a++; $a++; echo $a; → 3</property--></code></pre><code class="php"><br>
If the string contains something similar to 10чную SCH, we bring to the whole and incrementorum. The second increment gives "2e0" = 2*10^0 = 2. 2++ = 3.<br>
<br / >
the <pre><code class="php"><?php $a = "2d9"; $a+=2; echo $a; → 4</property--></code></pre><code class="php"><br>
Arithmetic increment without immediately causes the string to number, and we just believe 2+2.<br>
<br / >
Normal logic when you consider how PHP is used. It's like wondering why the increment short* increases the pointer value by 2, and not "how it should be" 1 :)</code></code>
If much interesting to write web sites in C and C++
In my projects, I use Apache HTTPD + interesting + modules content module written in C++ (although, from ++ there are only objects with inheritance, no templates, especially stl and boost, object semantics of the structures from C)
Apache HTTPD already has on Board a solid set of useful things (Google Apache Portable Runtime), deficiencies are corrected tons of good and accessible libraries (curl, imagemagick, openssl, anything from the world of C/C++, at least V8 + toc++ to generate bindingof and you have your scriptually javascripta server)
In addition, it is possible to obtain a memory management system from APR (memory pools), which eliminates severe headache with leaks, control of the lifetime resources and other pleasures.
For all this joy we have written a framework that gives to C++ code readability and development speed level interpreted languages, which the team already year wants finished to a full-fledged open Source project (here, only free time... but it looks like it will be in the next life).
The Assembly is built on a makefile, which is wonderful understands any IDE that is to Linux. Build incremental, as expected, compiling with the active work on a feature less than a second. Debugging with GDB (also perceived great IDE). Compile + deploy + run with one shortcut takes less than a second.
Working versions are rolled out via primitive CI (gitolite + hooks), the server collects himself, sets himself. Rolling out the same takes 5 to 10 seconds (on the server, rebuild completely, well, not the last role is assigned).
The web muzzle is going to buttrape (although recently coders tasted SemanticUI and was glad), the heart of a web muzzle - AngularJS.
But what about indexing, I will ask those familiar with SEO people? PhantomJS + a simple script to solve this problem. When it comes to the robot for a complete index of the server of course not very happy, but quite a stand and without noticeable brakes (0.1 MS vs 1.5 MS, with average ping of users in 150мс)
In General, you want to write websites in C is more than real, and sometimes more than efficient. Strong typing and the absence of which is interpreted at Runtime code solve a lot of problems with stability and security. And it's not in the standard level, and approach.