very nice stuff?!

I’ve always hardcoded the domain name in my htaccess’es, requiring me to make changes each time I deploy a new website.

The solution:

Behold, an alternate, generic method of redirecting non-www to www and www to non-www, requiring no changes between deployments!

Non-www to www

RewriteCond %{HTTP_HOST} !^www\.
RewriteRule (.*) http://www.%{HTTP_HOST}/$1 [R=301,L]

www to non-www

RewriteCond %{HTTP_HOST} ^www\.(.*)$ [NC]
RewriteRule ^/(.*)$ http://%1/$1 [R=301,L]

Bonus tip: Remove trailing slash from address line

RewriteRule ^(.+)/$ http://%{HTTP_HOST}/$1 [R=301,L]

I can hardly wait to get it FreeBSD 7.0 installed on it and setup a nice vhost system with Lighttpd + PostgreSQL. It’s gonna be awsome.
The server I’m currently using is ‘only’ Dual core XEON w/ 3.5Gb ram, and some random SATA drives, so the improvement should be huge

I’m quite excited, because its going to be my first server ever without Apache and MySQL installed. Or well, atleast until some stupid customer or what not (Biesbjerg?) demands it

If I can find the time, I will write a guide on how I’m going to configure my server (Lighttpd, PostgreSQL, Dovecot + sendmail, quota, fast-cgi php5), it might be useful for someone

Let me know if it has any interest to you guys

Specs

- Dell PE 1950 III
- PE1950 PCIE Riser (2 Slots) 1 S
- Quad Core Xeon E5410 (2.33GHz, 2x6MB, 1333MHz FSB) (SMP supported)
- 4GB (2x2GB Dual Rank DIMMs) 667MHz FBD (up to 32 GB)
- 3x 146GB SAS 10k 2.5″ HDPE1950 III 2.5″ HDD support chassis
- Perc 6i Integrated Controller
- 8X DVD-ROM Drive IDE
- PE1950 III Non-Redundant Power Supply No Power Cord
- TCP/IP Offload Engine 2P
- Broadcom TCP/IP Offload Engine functionality (TOE) Not Enabled
- PE1950 OpenManage kit and FI Driver
- CFI EMEA Service. (SV-Hi) OR PowerEdge, RAID/partition
- RAID 1 Single Container
- RAID 0 (1 HDD)
- Rack Rails Static Rapid 1 S
- PE1950 III – C2,MSSR0,ADD IN PERC5i/6i/SAS6iR, min 2 / max 2 or 4

/Jippi

I’m working on some security enhancements for my new website project, and in that progress, I have moved away from Apache2.2.3 and now using Lighttpd instead. Lighttpd does not use the same method of invoking php as apache does (mod_php vs. fast-cgi php).

In my eager to make the application as secure and locked down as possible, I went berserk in the php-cgi.ini file and turned on just about any security feature I could find (not safemode!), including the innocent looking switch called “open_basedir”.

That was rather late in the night, so when I resumed my work the following day, I have long forgotten about all the fancy security settings I had enabled in php.

A few hours ago I noticed that the complete render time for the front page suddenly was ~2seconds(!) – and comparing with apache’s load time for the same page (0.9s) I was quite disappointed at lighttpd and fastcgi, but refused to give up my new found love without a fair trial.

I attacked google with full force without finding any useful resources that might be able to explain my issue.. I was quite sure it was APC that did not function well in a fastcgi environment – and therefore my google madness was focused around the keywords “apc + fastcgi”

While chatting to PatrickDK in the #lighttpd channel @ freenode, it suddenly hit me – let’s try to disable everything I have “fixed” last night – starting with XDebug, some 3rd party php modules (syck, memcache, simplexml, fileinfo) – and then, open_basedir flag.

And with just one line commented out in my php-cgi.ini file, the render time went from 2.4 to 0.5 seconds !!(!!)

So just a little note to you guys out there, beware of the open_basedir setting in a setup where you include many files, it is a rather costly feature to use. And when running fastcgi php in chroot()ed setup, its not really required anyway

Its another +1 for lighty and fastcgi

And yes, I’m aware that open_basedir is slow because it has to validate EVERY file or path you work with inside php to check if its within the scope defined.

This time around it was more an impulze to buy some books, so I hope I will end up as satisfied as last time I purchased some books from them.

• Building Scalable Web Sites: Building, scaling, and optimizing the next generation of web applications
• RESTful Web Services
• High Performance Web Sites: Essential Knowledge for Front-End Engineers: Essential Knowledge for Front-end Engineers
• PostgreSQL (Developers Library)
• The CSS Anthology: 101 Essential Tips, Tricks and Hacks

I will probably give a short review of the books once I have read them

Preface

This time I have been busy playing with Lighttpd and mod_magnet. I found a blog post where darix mentions a post from sitepoint where they demostrate a technique to speed up the HTTP GET of javascript and CSS files.

Its quite simple really, instead of doing

<link href="/css/styles1.css" rel="stylesheet" type="text/css" />
<link href="/css/styles2.css" rel="stylesheet" type="text/css" />
<link href="/css/styles3.css" rel="stylesheet" type="text/css" />

you simply do

<link href="/css/styles1.css,styles2.css,styles3.css" rel="stylesheet" type="text/css" />

and then this script will automagic concat them together into one big file, and make lighty serve that one instead. (The same goes for javascript files, or anything really)

This could save ALOT of HTTP GET’s to your server, and increase the overall loadtime and performance of your site alot, especially if you use alot of Web2.0 stuff

However, the Lua source for their script was not made public, so nobody could benifit from their otherwise impressive speed gains.

Until now.. or, well, almost – its not their script, but my own attempt to mimic what they did in pure Lua

I tried to keep the dependencies down to a minimum, however I decided to use one, md5.
You can find the source here (5.1, 5.0)

Config

charset (Line 42)

Set it so it matches the headers in your lighttpd.conf file. Must match your mod_compress settings aswell to utilize that feature

prefix (Line 43)

Could be anything really, its just a mean to “namespace” your cached files so you can run the script with multiple configs

rootPath (Line 44)

The path to look for the files to concat, default should be fine for most DOC_ROOT/js and DOC_ROOT/css

concatRoot (Line 45)

The folder to store the concated files, can be anywhere

Setup

• Put the source below somewhere on your disk, I asume you name it bundle.lua
• Enable mod_magnet in lighttpd.conf
• Add magnet.attract-physical-path-to = ( “YOUR_PATH/bundle.lua” ) to your config where needed
• Restart lighttpd
• Modify your css/js links to utilize the new feature:)

Optimal you can enable mod_compress aswell, bundle.lua works out of the box with any other lighttpd module you may have

Source

download source

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87

--- get stat information on a path
-- @param path to stat
-- @param the ftype to return
-- @return false if ftype or path does not exist
function file_info(path, ftype)
    local attr = lighty.stat(path)
    if attr and attr[ftype] then
        return attr[ftype]
    end
    return false
end
 
--- Wrapper for reading a full file into a string
-- @param filename Full path to the file
-- @return a string with the content of the file
function read_file(filename) 
    local content = ""
    if file_info(filename, "is_file") then
        local file = io.open(filename, "r")
        content = file:read("*a")
        io.close(file)
    end
    return content
end
 
--- Wrapper for writeing content to a file
-- @param filename Full path to the destionation file
-- @param content The string to write
function write_cache(filename, content) 
    local file = io.open(filename, "w")
    file:write(content)
    io.close(file)
end
 
--- Concat multiple files into one file 
-- @param lighty lighty global variable passed to the method
-- @param match The files that will be concat into a file
-- @param fileExtension Do !NOT! include the dot ( . )
function combine_files(lighty, files, fileExtension)
    require "md5" 
 
    local charset       = "; charset=utf-8"
    local prefix        = "cache-"
    local rootPath      = lighty.env["physical.doc-root"] .. fileExtension .. "/"
    local concatRoot     = "/tmp/cache/"
    local lastModified  = 0
 
    for file in string.gmatch(files, "(.-\." .. fileExtension .. "),?") do
        local fullPath = rootPath .. file
        modTime = file_info(fullPath, "st_mtime")
        if type(modTime) == "number" then
            lastModified = math.max(lastModified, modTime)
        end            
     end
 
     local hash = lastModified .. "-" .. md5.sumhexa(files)
     lighty.header["Etag"] = '"' .. hash ..'"'
 
    local cacheFile = prefix .. hash .. '.' .. fileExtension
    if not file_info(concatRoot .. cacheFile, "is_file") then
        local content = ""
        for file in string.gmatch(match, "(.-\." .. fileExtension .. "),?") do
            content = content .. "\n"
            content = content .. "/**\n"
            content = content .. " *\n"
            content = content .. " *  " .. file .. "\n"
            content = content .. " *\n"
            content = content .. " */\n"
            content = content .. "\n"
            content = content .. read_file(rootPath .. file)
        end
        write_cache(concatRoot .. cacheFile, content )
    end
    lighty.env["physical.path"] = concatRoot .. cacheFile
end
 
if (not file_info(lighty.env["physical.path"], "is_file")) then
    css = string.match(lighty.env["physical.path"], "css/(.*\.css)")
	if css then
        return combine_files(lighty, css, "css")
    end
 
    js = string.match(lighty.env["physical.path"], "js/(.*\.js)")	    
	if js then
        return combine_files(lighty, js, "js")
	end 
end

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51

$data = array( 
    1 => array('Post' => array( 
            'id' => 1, 'post_comment_count' => 2, 
            'name' => 'My first title', 'body' => 'My first body' 
        ), 
        'PostComment' => array( 
            array( 
                'id' => 1, 'post_id' => 1, 'is_active' => 1, 
                'name' => 'My first comment', 'body' => 'My first comment'), 
            array( 
                'id' => 2, 'post_id' => 2, 'is_active' => 1, 
                'name' => 'My second comment', 'body' => 'My second comment'), 
            array( 
                'id' => 3, 'post_id' => 2, 'is_active' => 1, 
                'name' => 'My third comment', 'body' => 'My third comment') 
        ) 
    ), 
    2 => array('Post' => array( 
            'id' => 2, 'post_comment_count' => 1, 
            'name' => 'My second title', 'body' => 'My second title' 
        ), 
        'PostComment' => array( 
            array( 
                'id' => 4, 'post_id' => 2, 'is_active' => 
                1, 'name' => 'My fourth comment', 'body' => 'My fourth comment' 
            ) 
        ) 
    ), 
    3 => array('Post' => array( 
            'id' => 3, 'post_comment_count' => 0, 
            'name' => 'My third title', 'body' => 'My third title' 
        ), 
        'PostComment' => array(                   
 
        ) 
    ) 
);                   
 
$dataExtra = array( 
    4 => array('Post' => array( 
            'id' => 4, 'post_comment_count' => 1, 
            'name' => 'My fourth title', 'body' => 'My fourth title' 
        ), 
        'PostComment' => array( 
            array( 
                'id' => 5, 'post_id' => 4, 'is_active' => 1, 
                'name' => 'My fifth comment', 'bdoy' => 'My fifth comment' 
            ) 
        ) 
    ) 
);

Set::merge

PHP doc blcok

This function can be thought of as a hybrid between PHP's array_merge and
array_merge_recursive. The difference to the two is that if an array key
contains another array then the function behaves recursive (unlike array_merge)
but does not do if for keys containing strings (unlike array_merge_recursive).
See the unit test for more information.         

Note: This function will work with an unlimited amount of arguments and
typecasts non-array parameters into arrays.

A bit like array_merge_recursive. Merge joins two arrays on their keys.
If we want to add the $extraData array into the $dat most people would probably think, hey, lets merge them using Set::merge… but.. no, dont, unless you know what your doing.

1

pr(Set::merge($data, $dataExtra));

The above cod will only produce the ‘expected’(?) result because I have added the numeric key 4 in the $extraData array (Line 40). If I had, as I did the first time around, just let php auto-increment the keys for me, both $datad and $extraData would contain the numeric index key 0, and since Set::merge works on keys, $extraData’s Fourth post would overwrite $data’s First post where they had anything in common (id, name, post_comment_count, body and everything in the first PostComment (My first comment)).

In case I had let php handle the index keys, I would have to use Set::insert to append the fourth comment to the list.

A better use of Set::merge is in CAKE/libs/model/model.php where is automagic merges ‘actsAs’ from your current model (PostComment) and your AppModel, so you automagic inherit the global behaviors from AppModel:

384
385
386
387
388
389
390
391
392
393
394
395
396
397
398

// model.php 
if (is_subclass_of($this, 'AppModel')) { 
    $appVars = get_class_vars('AppModel'); 
    $merge = array();             
 
    if ($this->actsAs !== null || $this->actsAs !== false) { 
        $merge[] = 'actsAs'; 
    }             
 
    foreach ($merge as $var) { 
        if (isset($appVars[$var]) && !empty($appVars[$var]) && is_array($this->{$var})) { 
            $this->{$var} = Set::merge($appVars[$var], $this->{$var}); 
        } 
    } 
}

A funny thing is that Set::merge is still rarely used in the core code yet. Probably because array_merge (or am()) is faster for simple operations like just merging two simple 1 level arrays like a config array.

Set::filter

PHP doc blcok

Filters empty elements out of a route array, excluding '0'.

This method is not recursive. So doing

1
2

< ?php 
pr(Set::filter($data, true));

wont change anything in the array sinc no elements in first level is empty.

1
2
3
4
5
6
7
8
9

pr(Set::filter($data[3], true));      
 
array('Post' => array( 
    'id' => 3, 
    'post_comment_count' => 0, 
    'name' => 'My third title', 
    'body' => 'My third title' 
    ) 
);

We will notice that the PostComment array is gone, since its empty

Set::pushDiff

PHP doc blcok

Pushes the differences in $array2 onto the end of $array

That sounds simple enough, and it really is.

1

pr(Set::pushDiff($data, $dataExtra));

just pushes the $dataExtra on the end of $data, since they have nothing in common. Again, this method operates on the array keys when comparing arrays.

Set::map

PHP doc block

Maps the contents of the Set object to an object hierarchy.
Maintains numeric keys as arrays of objects

That sounds very fancy, and it is! A lot of people migrating to CakePHP from Symfony (PHPDoctrine, Propel) will probably be confused at first that all model data in CakePHP is arrays, and not objects like other frameworks. This method attemps to ‘fix’ this issue by converting a model data array to a set of objects (StdClass). StdClass is a build-in dummy class in PHP, and it doesnt really do anything but contain data, no methods or fancy magic there.

1
2
3
4
5
6
7
8
9

< ?php 
// Convert to objects 
$map = Set::map($data); 
// id of first post (Object style) 
echo $map[0]->id; 
// name of first comment to first post (Object style) 
echo $map[0]->PostComment[0]->name; 
// name of first comment to first post (array style) 
echo $map[0]['PostComment'][0]['name']; // WONT WORK, Cannot access StdClass as an array

As you can see above, its important to decide what method you want to use, Object or Array style, since they cannot be mixed.
Also, the return from Set::map is a mix of arrays and StdClass, thats why I can access $map[0] as array, but not access $map[0]['name'].

Set::numeric

PHP doc blcok

Checks to see if all the values in the array are numeric

Not much to say on this one, because it doesnt do more or less than the name implies.

1
2
3
4
5
6
7
8

// false 
var_dump(Set::numeric($data)); 
// false 
var_dump(Set::numeric($data[1])); 
// false 
var_dump(Set::numeric($data[1]['PostComment'])); 
// true 
var_dump(Set::numeric(array(1,2,3)));

——-

Thats it for now, next time I will look at:

Set::enum, Set::format, Set::extract, Set::insert, Set::remove, Set::check, Set::diff, Set::isEqual, Set::contains, Set::countDim, Set::normalize, Set::combine and Set::reverse.

Look forward to it

I have made a small change to the feeds now. Everybody who has subscribed by Feedburner should not notice anything strange.

The reason for the change is that I want to be able to blog in Danish too once in a while, and to avoid the ‘danish pølsesnak’ for you guys who does not understand Danish, I have moved the feed location to include only posts with the ‘english’ tag in the future

English posts only (Hopefully your using this one right now ), Danish posts only and Danish+English posts

/Jippi

Its actually a quite good presentation, too bad he ran out of time in the end