summaryrefslogtreecommitdiff
path: root/doc/cml.txt
blob: ae6906783c38fcd2b577d2a988ee87967294ea27 (plain)
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
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
=========================
CML (Cache Meta Language)
=========================

---------------
Module: mod_cml
---------------

:Author: Jan Kneschke
:Date: $Date: 2004/11/03 22:26:05 $
:Revision: $Revision: 1.2 $

:abstract:
  CML is a Meta language to describe the dependencies of a page at one side and building a page from its fragments on the other side using LUA.
  
.. meta::
  :keywords: lighttpd, cml, lua
  
.. contents:: Table of Contents

Description
===========

CML (Cache Meta Language) wants to solves several problems:

 * dynamic content needs caching to perform
 * checking if the content is dirty inside of the application is usually more expensive than sending out the cached data
 * a dynamic page is usually fragmented and the fragments have different livetimes
 * the different fragements can be cached independently

Cache Decision
--------------

A simple example should show how to a content caching the very simple way in PHP. 

jan.kneschke.de has a very simple design:

 * the layout is taken from a template in templates/jk.tmpl
 * the menu is generated from a menu.csv file
 * the content is coming from files on the local directory named content-1, content-2 and so on

The page content is static as long non of the those tree items changes. A change in the layout 
is affecting all pages, a change of menu.csv too, a change of content-x file only affects the
cached page itself.

If we model this in PHP we get: ::

  <?php

  ## ... fetch all content-* files into $content
  $cachefile = "/cache/dir/to/cached-content";

  function is_cachable($content, $cachefile) {
    if (!file_exists($cachefile)) {
      return 0;
    } else {
      $cachemtime = filemtime($cachefile);
    }

    foreach($content as $k => $v) {
      if (isset($v["file"]) && 
          filemtime($v["file"]) > $cachemtime) {
        return 0;
      }
    }

    if (filemtime("/menu/menu.csv") > $cachemtime) {
      return 0;
    }
    if (filemtime("/templates/jk.tmpl") > $cachemtime) {
      return 0;
    }
  }
  
  if (is_cachable(...), $cachefile) {
    readfile($cachefile);
    exit();
  } else {
    # generate content and write it to $cachefile
  }
  ?>

Quite simple. No magic involved. If the one of the files is new than the cached
content, the content is dirty and has to be regenerated. 

Now let take a look at the numbers:

 * 150 req/s for a Cache-Hit
 * 100 req/s for a Cache-Miss

As you can see the increase is not as good as it could be. The main reason as the overhead
of the PHP interpreter to start up (a byte-code cache has been used here).

Moving these decisions out of the PHP script into a server module will remove the need
to start PHP for a cache-hit. 

To transform this example into a CML you need 'index.cml' in the list of indexfiles 
and the following index.cml file: ::

  output_contenttype = "text/html"

  b = request["DOCUMENT_ROOT"]
  cwd = request["CWD"]

  output_include = { b .. "_cache.html" }

  trigger_handler = "index.php"

  if file_mtime(b .. "../lib/php/menu.csv") > file_mtime(cwd .. "_cache.html") or
     file_mtime(b .. "templates/jk.tmpl")   > file_mtime(cwd .. "_cache.html") or
     file_mtime(b .. "content.html")        > file_mtime(cwd .. "_cache.html") then
     return CACHE_MISS
  else 
     return CACHE_HIT
  end

Numbers again:

 * 4900 req/s for Cache-Hit
 *  100 req/s for Cache-Miss

Content Assembling
------------------

Sometimes the different fragment are already generated externally. You have to cat them together: ::

  <?php
  readfile("head.html");
  readfile("menu.html");
  readfile("spacer.html");
  readfile("db-content.html");
  readfile("spacer2.html");
  readfile("news.html");
  readfile("footer.html");
  ?> 

We we can do the same several times faster directly in the webserver. 

Don't forget: Webserver are built to send out static content, that is what they can do best.

The index.cml for this looks like: ::

  output_contenttype = "text/html"
  
  cwd = request["CWD"]
   
  output_include = { cwd .. "head.html", 
                     cwd .. "menu.html",
                     cwd .. "spacer.html",
                     cwd .. "db-content.html",
                     cwd .. "spacer2.html",
                     cwd .. "news.html",
                     cwd .. "footer.html" }
  
  return CACHE_HIT

Now we get about 10000 req/s instead of 600 req/s.

Power Magnet
------------

Next to all the features about Cache Decisions CML can do more. Starting
with lighttpd 1.4.9 a power-magnet was added which attracts each request
and allows you to manipulate the request for your needs.  

We want to display a maintainance page by putting a file in a specified
place:

We enable the power magnet: ::

  cml.power-magnet  = "/home/www/power-magnet.cml"

and create /home/www/power-magnet.cml with: ::

  dr = request["DOCUMENT_ROOT"]

  if file_isreg(dr .. 'maintainance.html') then
    output_include = { 'maintainance.html' }
    return CACHE_HIT
  end

  return CACHE_MISS

For each requested file the /home/www/power-magnet.cml is executed which
checks if maintainance.html exists in the docroot and displays it
instead of handling the usual request. 

Another example, create thumbnail for requested image and serve it instead
of sending the big image: ::

  ## image-url is /album/baltic_winter_2005.jpg
  ## no params -> 640x480 is served
  ## /album/baltic_winter_2005.jpg/orig for full size
  ## /album/baltic_winter_2005.jpg/thumb for thumbnail

  dr = request["DOCUMENT_ROOT"]
  sn = request["SCRIPT_NAME"]

  ## to be continued :) ... 

  trigger_handler = '/gen_image.php'

  return CACHE_MISS


Installation
============

You need `lua <http://www.lua.org/>`_ and should install `libmemcache-1.3.x <http://people.freebsd.org/~seanc/libmemcache/>`_ and have to configure lighttpd with: ::

  ./configure ... --with-lua --with-memcache

To use the plugin you have to load it: ::

   server.modules = ( ..., "mod_cml", ... )

Options
=======

:cml.extension: 
  the file extension that is bound to the cml-module
:cml.memcache-hosts:
  hosts for the memcache.* functions
:cml.memcache-namespace:
  (not used yet)
:cml.power-magnet:
  a cml file that is executed for each request

Language
========

The language used for CML is provided by `LUA <http://www.lua.org/>`_. 

Additionally to the functions provided by lua mod_cml provides: ::

  tables:

  request
    - REQUEST_URI
    - SCRIPT_NAME
    - SCRIPT_FILENAME
    - DOCUMENT_ROOT
    - PATH_INFO
    - CWD
    - BASEURI

  get
    - parameters from the query-string

  functions:
  string md5(string)
  number file_mtime(string)
  string memcache_get_string(string)
  number memcache_get_long(string)
  boolean memcache_exists(string) 


What ever your script does, it has to return either CACHE_HIT or CACHE_MISS. 
It case a error occures check the error-log, the user will get a error 500. If you don't like 
the standard error-page use ``server.errorfile-prefix``.