Skip to content

Home · Changes

Page history
thimo created page: home authored by Thimo Kraemer's avatar Thimo Kraemer
Hide whitespace changes
Inline Side-by-side
home.md 0 → 100644
View page @ 27a469e1
Introduction
============
KSS is an acronym for Kinetic Style Sheets. KSS is a client side
framework for implementing rich user interfaces with AJAX
functionality.
In this document, we assume the reader understands AJAX and its
purpose: to make pages in a browser behave more like a desktop
application. A primary AJAX characteristic
is that, instead of the user loading or reloading a web page, the web client
can contact the server and recieve information from it, then can change the
page's content or appearance without leaving it. As a result, the user
experiences faster interactions and a more "desktop"-like interface.
Plus, several tasks can be implemented with AJAX that are normally not
possible to achieve with static HTML pages.
The intention was not to create another JavaScript
framework for AJAX. Rather, we wanted to reuse existing libraries.
However, we also identified a few major problems in our everyday life
as AJAX developers, and by creating KSS we are attempting to offer
better solutions.
In many use cases, the "thin client" approach provides the best
results. Instead of performing complex DOM
modifications from JavaScript, the client consults the server, and the
server sends back an HTML page which is simply used by the client to
replace a part of the page.
Developing a "fat client" should seldom be necessary. But for the
use cases when fat clients are desirable, the process should work in
a reusable way, by encapsulation of plugins.
Developers have different skills and interests: site design,
application programming, plug-in creation, and so forth. Designers,
for example, are largely interested in tools that help them style a
site's look-and-feel, while programmers want to focus on
implementing that user interface with responsive, desktop-like
efficiency. As a result,
throughout this documention we highlight information peculiar to the
interests of several types of KSS user:
- KSS designer. A completed application can be bound to a page by writing
a KSS resource file. This task can be achieved by a
designer or by a programmer without any JavaScript knowledge. The
similarity of the KSS format to CSS assures the stylesheet is
developer friendly.
- KSS application developer. A developer can create an
application by reusing several KSS "plugins" that contain the
JavaScript code necessary for most client tasks. A developer
needs to write KSS resource files, and also must create server
side methods that command the client. This activity requires
you to know server side programming, but
JavaScript knowledge is not required.
- KSS plugin developer. Creating a plugin is the most complex but least often
needed activity. Plug-in development requires an extensive
knowledge of both browser side JavaScript programming and the
browsers themselves. Knowledge of the KSS plugin API is also
needed.
In this document, we assume you are familiar with HTML and
how web pages are created. We also assume that you know CSS, and, in
particular, that you understand CSS selectors. We do
not require you to be a Javascript expert to understand this
document.
A typical KSS application has the following characteristics:
- There is no JavaScript code or event binding in the HTML itself
(although we include JavaScript that contains the necessary client
side software for KSS to work).
- The behaviour is defined by a KSS resource file. A resource file has a syntax
similar to CSS, and its functionality is similar too. But instead of
being a static stylesheet, a KSS resource file defines how the
browser events should be bound to the HTML page.
- The client calls server side code, called "server actions," if certain browser
events are triggered (such as a user clicking on a node).
This server side code assembles a response that
consists of "kss commands" which are sent back to the client and which execute the
required behaviour.
Our goal in KSS to avoid the need to write JavaScript. JavaScript
is still necessary for some use cases, but even such
circumstance, we advise you choose JavaScript only as a last
choice. That is, first develop a "thin client" solution, and turn
it in to a "fat client" later, when you can't find another way to
solve the programming problem or you decide to spend excess time
on optimization. Besides, this gives you a possibility to
deliver a working application, and thus to learn the application
domain sooner. But often, it turns out that the thin client is
"good enough" and the development of the fat client can be
omitted altogether.
Architectural overview
======================
In this section, we provide a high level overview of KSS' system
architecture.
The principal characteristics of the KSS process are:
1. Process control is embedded into the KSS resource file
(including how to bind browser events to the plugins, and how to
dispatch them to the server), and server action code (that decides
what commands to marshal in return). Thus, application control logic
is not implemented on the client side, but is the
responsibility of the server.
2. All JavaScript code (apart from the KSS core) is a KSS
plugin, in most cases an event binder plugin or a client
action plugin. Apart from this, no JavaScript code appears in
the system. In particular, no JavaScript
ever appears in the HTML pages.
![1-kss-architecture](/uploads/b6560aee9bf788a1fbab959c7575d6b9/1-kss-architecture.png)
This diagram shows the architecture of KSS. Magenta and yellow
identify the two different types of plugins, and red highlights
the controlling information that governs the actual AJAX
functionality.
A major intent of the architecture of KSS is to separate design
from implementation.
KSS provides the execution flow similar to a simple programming language,
however yet does not have a "program." A command language can
execute a set of parametrized commands.
You can start implementing a thin client solution by using the
built-in set of KSS plugins. If it becomes necessary, you can
gradually fatten your client by developing new custom plugins.
Doing the latter naturally requires JavaScript programming, but
even so the result should be a well-componentized implementation
wherein
the details are totally decoupled from the level of the skin design.
To understand the entire KSS process, contemplate the complete flow of events
of an AJAX action. First, all actions are triggered from the client browser by
browser events. These are the same events that a programmer normally would bind
to JavaScript functions. Some of these events are generated by the user's
interaction (such as clicking on the page) or are triggered in some other
way (e.g. timer events).
Normally, JavaScript code
on the page would handle these events; in KSS, however, the events are
handled by event binder plugins. These plugins control both how the
event is bound and the actual process that executes when the event is
triggered.
Several event binder plugins are built into the KSS system. The
simplest plugin binds to a browser event via a
CSS selector. Other event type plugins are built-in. It is also
possible to extend the system by writing a custom event binder
plugin.
Such event binder plugins can bind to additional browser events and perform
more complex functionality. The way to bind these event plugins to browser
events is specified by the KSS resource file, a static resource file that
accompanies the HTML page. The format of this resource is called KSS which is
similar to CSS in syntax and nature; however, while CSS specifies
the static style of the content, KSS binds the dynamic behaviour
to a page.
The binding code for the event binder plugins runs at page
loading time. Event binder plugins may also specify the code to
be executed when the event is triggered, but usually they pass
execution to the event rule action dispatcher. The dispatcher
uses the KSS rule to decide what action to take. When
the action is a server action (the most common), control is
passed to the server. But client actions can also execute locally.
It is also possible to execute multiple actions for the same event.
If an event triggers a server action, control is passed through
an RPC service or direct URL request, which sends
events in the form of XMLHttpRequests to the server.
Once the XMLHttpRequests are sent, they arrive on the server as
RPC or normal URL requests. One server action is designated to reply to each
request. The methods do whatever their designated task is, then command the
client to do something in return.
This is a principal point of KSS: at this point in the process the server actions
decide what should happen on the client. The server side event
handlers do this by assembling a sequence of commands with
parameters. This command calling sequence is sent back to the
client as a response.
The commands consist of a selector (typically) and a client
action that is called with parameters. When the command
marshaller receives this sequence of commands, it looks up the
nodes identified by the selector. The command marshaller calls
each assigned client action in the required order,
with the parameters supplied by the server. The client actions
continue the flow of execution on the client side; this may
result in manipulating the DOM content of the page, binding
further events, and as a consequence, the execution of further
actions. In the end, the change is visible for the user,
finishing the entire AJAX cycle.
The client actions, like the event binders, ship as "core" plugins
into the KSS system. The simplest and also most useful client action
(replaceInnerHTML) replaces a selected tag in the DOM with a
string value supplied by the server as a parameter of the command.
This covers the most important KSS use case: in response to a user
action, the browser replaces part of the page that is
completely rendered on the server side. This results in a thin
browser client where most tasks are delegated to the server;
in many AJAX use cases, this is also the proper solution.
However, additional
local actions can carry out other common tasks, and you can
extend the system for your needs with very complex actions. In
the more complex cases, you may even want the server to return a
set of data to be turned into more complex DOM manipulation by
the accompanying command plugin.
Syntax and semantics of the KSS resource file
=============================================
In this section we detail what a KSS resource file contains
and how to use it.
At-config rule
--------------
The main configuration is done within an @config rule:
```css
@config {
protocol: json-rpc;
date-encoding: iso8601;
timeout: 5000;
endpoint: /path/to/service;
}
```
protocol
The protocol to use. A value of "json-rpc" (default), "json-rpc-v1",
"json-rpc-v2" (synonym for "json-rpc"), "xml-rpc" or "url-encoded".
The latter can be used if there is no RPC service available.
date-encoding
How dates should be encoded. A value of "iso8601" (default),
"@timestamp@", "class-hinting" or "asp.net".
timeout
The timeout for server actions (defaults to 10000).
endpoint
The endpoint url to the RPC service.
Simple cases
------------
KSS is a resource format similar to CSS. It contains "rules." A rule
selects a HTML element from the page, binds it to an event, and
specifies the "actions" to happen when this event occurs.
A simple example is this:
```css
#title_save:click {
kss-action-server: saveTitle;
}
```
This binds the node selected by the CSS selector ``#title_save`` to the
``click`` event. If this node is clicked, we want an action to happen.
In this case, this is a "server action" executed on the server. A server
action must be the name of a remote RPC method.
Another example:
```css
#sportlet-main:timeout {
evt-timeout-delay: 5000;
kss-action-server: refreshPortlet;
}
```
This binds the ``timeout`` event to the HTML node selected by the
CSS selector ``#portlet-main``. The ``evt-timeout-delay`` specifies a
parameter used during the binding of the event. In this example,
it specifies how often the timeout event should poll. Similar to
the previous case, we also execute a server action here. The
method ``refreshPortlet`` is called every five seconds, and
results in refreshing a portlet in a page.
Complex cases
-------------
The goal is to enable the creation of complex event types with
abstract events that are not equivalent to browser events. A
simple example is ``timeout``: this is not a
browser event, but it is relatively
simple to bind it. In even more complex cases, the plugin writer can freely
implement how the events should be bound. The key point to keep
in mind is that once the events are triggered, we handle them
within the system, as if they were normal browser events.
The ``timeout`` event is contained in the "core" plugins of KSS.
The KSS stylesheet rules resemble CSS rules, but
we attach a KSS event name at the end to designate the event
to be bound to the nodes selected by CSS. The event
name is prefixed with a colon.
The attributes inside the rule also follow a syntax very similar
to CSS, but with different semantics. We
explain more about the parametrizing semantics a bit later.
Overall, however, the rule contains different properties that
define:
- parameters for the event binding
- server and client actions, with parameters (there can be more of these)
A rule follows the following syntax:
```css
[<CSS> <CSS> <CSS> ...] <CSS_SELECTOR>:<KSS_EVENT_NAME> {
/* event parameters */
...
/* actions (server or client) */
...
}
```
Consider the following example:
```css
div#thisnode a:timeout {
/* event parameters */
evt-timeout-delay: 3000;
/* actions */
kss-action-server: updateInfo;
remark: 'Updating from timeout';
color: red;
kss-action-client: log;
message: 'Updating from timeout';
}
```
A ``timeout`` event is bound; in this case the ``div#thisnode a`` CSS
selector selects any nodes. The event is bound to the selected nodes, but in
case of the timeout event the node itself does not matter.
The parameter ``delay=3000`` specifies a 3000ms
timeout recurring tick. When the events are executed, two actions
(updateInfo and log) are executed, each one with the supplied
parameters. The remote action is executed on the server, and then
sends back its command response to the client. The "log" action
is a special client action that is
implemented as a JavaScript action plugin; this
particular one gives a log message.
Event selectors
---------------
The event selector follows the css selector with a colon in
the KSS rule:
:eventname
Examples:
:click
:timeout
The full event rule
-------------------
The name of the event follow the CSS selector after a colon:
```css
div#recent-portlet:click { ... }
div.banner:timeout { ... }
```
The event name in KSS must be preceded by a normal CSS selector.
It cannot stand by itself.
A full example for an event method rule is provided later in this document.
Specifying parameters
---------------------
The general schema for the parameter specification:
```css
.... {
evt[-<eventname>]-preventdefault: [true|false];
evt[-<eventname>]-preventbubbling: [true|false];
evt[-<eventname>]-allowbubbling: [true|false];
evt-<eventname>-<key>: <value>;
kss-selector: <selector>;
kss-precondition: [true|false];
...
kss-action-[server|client]: <actionname>;
kss-selector: <selector>;
kss-precondition: [true|false];
kss-includeform: [<form_name>|@currentform];
<key1>: <value1>;
<key2>: <value2>;
...
kss-action-[server|client]: <actionname>;
kss-selector: <selector>;
kss-precondition: [true|false];
kss-includeform: [<form_name>|@currentform];
<key1>: <value1>;
<key2>: <value2>;
...
}
```
The keys and action names can be in camelcase (CamelCase). Although
the original CSS disallows the usage of uppercase characters in the
identifier, they are allowed in KSS.
"Event parameters" mean that these are parameters to the event
binding itself. There are different possibilities to declare them:
```css
evt-preventdefault: true;
evt-preventdefault: false;
evt-click-preventdefault: true;
evt-submit-preventdefault: true;
```
Declarations with omitted event names are applied to all event types
of the selector (that could be more than one).
Event actions can be defined with their parameters. An event
action must be a server or a client action. It is possible to
have more than one event action within a rule:
```css
#thisnode:timeout {
kss-action-server: updateInfo;
remark: 'Updating from timeout';
color: red;
kss-action-client: log;
message: 'Updating from timeout';
}
```
Parameter producer functions
----------------------------
A value simply represents a value. Quotes are optional, but as seen
in this example, both single and double quotes can be used also:
```css
... {
size: 12pt;
typeface: "Bitstream Vera Sans";
typeface: 'Bitstream Vera Sans';
}
```
It is possible to create extensions to acquire a parameter in another
way as a constant; these are called parameter producer functions.
When calculating a value, these preset functions can take the
event rules, the state of the event, and the page in
consideration. We show some examples here, and we will give the
complete list of the selectors, too, later:
```css
... {
node_id: nodeAttr(id);
rownum: dataAttr("rownum", true);
}
```
The parameter producer functions operate on the scope of the
node which triggered the current event.
KSS action parameters
---------------------
KSS action parameters are special predefined parameters. Strictly
speaking, they are not really freely usable parameters to specify
a value for an event; rather, they modify the way the (client or
server) action works.
The action parameters are distinguishable from normal parameters
by their name: they have the KSS prefix. (Consequently, normal
parameters are not allowed to start with this prefix.)
Only a few action parameters are implemented, so we present them
all here:
### kss-selector
``kss-selector`` modifies the scope
of the action. The default execution scope is the same node where
the event was triggered. With ``kss-selector``, we can execute
the action on a different or multiple nodes:
```css
... {
kss-action-client: actionName;
kss-selector: selector;
}
```
### kss-precondition
``kss-precondition`` prevents the execution of all declared actions
in a rule if its value evaluates to ``false``:
```css
... {
kss-action-server: actionName;
kss-precondition: confirm('Are you sure?');
}
```
### kss-includeform
``kss-includeform`` transparently adds an entire
form to the parameters of a KSS action. It is global to the
action: all field variables of the form are added with their
name as they appear in the form:
```css
... {
kss-action-server: actionName;
kss-includeform: formname;
}
```
The value of the declaration can specify to submit a given form
or the current form (in which the event has been triggered).
Variations can be seen in the next example:
```css
kss-includeform: formname;
kss-includeform: @currentform;
```
Summarizing the rule definition basics
--------------------------------------
A complete example rule follows here:
```css
div#portlet-recent:timeout {
evt-timeout-delay: 2000;
kss-action-server: replaceMacro;
selector: #portlet-recent;
macropath: portlet_recent/macros/portlet;
}
```
Here, ``delay="2000"`` is used as a parameter for the binding of the
timeout event, and ``selector="#portlet-recent"``,
``macropath="portlet_recent/macros/portlet"`` are used for calling
the "replaceMacro" method. In this example, ``selector`` is not a
kss action property; it is a normal parameter passed to the
server.
Another example:
```css
div.menu-item:load {
kss-action-server: replaceWithRenderedText;
text: textContent();
size: 12pt;
typeface: "Bitstream Vera Sans";
}
```
How rule merging is done
------------------------
In contrast to CSS and the original KSS implementation, in KSS-RPC
there is no rule merging done at all. So a declaration of two rules
in the context of the same node results in the binding of two events.
Error handlers
--------------
It is possible to attach error handlers to server actions.
This is implemented as an additional event type ``error``:
```css
body:error {
kss-action-client: alert;
message: errorAttr('message');
}
```
The error actions can accept the special parameters provider ``errorAttr``
to access properties of the current error object.
In practice, the main reasons that an error will occur during a remote action
execution are server errors or timeouts. It is also possible
for a server method to raise an exception on purpose, with the expectation that
the error handler will address the case on the client side in a
designated way.
Action names
------------
Action names are declared following the kss-action-client or
kss-action-server keys in the rule definition.
An action name is the name of either a server or client action,
depending if kss-action-client or kss-action-server is applied. These are
the two choices we have: pass the control flow to the server
(the most common case), or handle an action locally on the
client (in which case, a JavaScript plugin will be called).
If an action key is missing from an event or method rule, no
action is carried out.
A rule definition can have multiple client and server actions
defined. Actions are executed in the declared order but server
actions are executed asynchronously.
Getting information from the DOM (parameter providers)
======================================================
Actions must be executed with parameters. Although static
parameters can be sufficient for some application, most of the
time a program expects to fetch some information from the DOM
(the page where the event executing the action has triggered).
This pattern is solved by the "parameter providers" that we
introduced earlier in chapter 1.3.
For those not familiar with the term, we mention that DOM stands
for the Document Object Model. This refers to the internal
representation of our HTML page. So when talking about the DOM,
we mean accessing information from our page from a program or as in
this case, from KSS. And this is the exact purpose of the
KSS parameter providers.
A parameter provider looks like a function with a number of
parameters, and can stand in place of the value of a parameter:
```css
... {
kss-action-server: doIt;
whatisit: parmProvider(arg1, arg2, ... argN);
}
```
There are two standard examples used most often.
In the first case, we want to take an attribute of a node.
HTML:
```html
<div id='our-node'>The content</div>
```
KSS:
```css
div#our-node:click {
kss-action-server: doIt;
id: nodeAttr('id');
}
```
We won't examine this line-by-line, at this point, but the important
piece to look at is the parameter 'id' that is passed with the value
'our-node'.
In the other use case, we want to include special markup with our
HTML, but we can not use HTML attributes for that. Existing HTML
attributes already have a specific purpose; using an attribute
not contained in the HTML standard would result in non-validating
pages. To avoid this problem, we use the HTML5 data-* attributes
to achieve the same result. These attributes can be acquired by
the ``dataAttr`` parameter provider.
HTML:
```html
<div id='our-node' data-marker='marker-id'>The content</div>
```
KSS:
```css
div#our-node:click {
kss-action-server: doIt;
marker: dataAttr('marker');
}
```
The value 'marker-id' is marshalled to the doIt action. We can
use any variable name, since they all become attributes within
the data-* attributes.
It is a further enhancement of this use case to acquire these
attributes in a recursive way. For example, let us consider
a widget we created. Using KSS attributes recursively lets us
mark up an entire widget with a single KSS attribute, and we can
access the value of the single attribute from ``dataAttr`` if an event
happens anywhere inside the widget.
HTML:
```html
<div id='our-node' data-widgetid="widget2000">
<a class='link' href='firstpage.htm'>First page</a>
<a class='link' href='secondpage.htm'>Second page</a>
</div>
```
KSS:
```css
a.link:click {
kss-action-server: doIt;
href: nodeAttr(href);
widgetid: dataAttr('widgetid', true);
}
```
The second parameter in ``dataAttr('widgetid', true)`` tells KSS to look
further in the parents of the node, until it finds the attribute.
Introduction to commands
========================
"KSS commands" serve a very important role in a KSS application.
They enable the application to use server-side code to instruct
what the browser should do on the client side. A command
specifies:
- A client action that the command will execute. This action
stands with parameters that travel in the command from the
server to the client.
- In the majority of commands this is accompanied by a "selector"
parameter which identifies the node or nodes on which the action needs
to be executed. If there is no selector specified, the action is
executed in the context of the node that triggered the event.
Several commands are preimplemented in the KSS plugins.
Practically, all the client actions can be executed both
from the client and from the server.
When a server action executes, it assembles a response which it
will send back to the client. This response consists of a set of
commands.
A command instructs KSS on the browser to execute a client action
combined with a selector that is applied to select nodes for the
execution. This means that the selector is executed first on the
page, and the client action is executed on the resulting nodes.
The execution may occur on zero or more nodes. Executing on zero
nodes is permitted and results in no error condition.
Commands are constructed on the server side. To demonstrate a complete
example, let's examine an example. Look at the doIt action:
```python
from kss import KSSResponse
@rpcmethod
def doIt(widgetid, href):
kss = KSSResponse()
kss.replaceInnerHTML('div#result',
html='<p>Parameters: %s, %s</p>' % (widgetid, href))
return kss.response
```
The doIt action, as implemented above, commits a single command
``replaceInnerHTML`` that causes the target node to be replaced
by the value of the single parameter 'html' of the action.
The commands, as the response given by the server action, are
returned in RPC format (either JSON or XML). This is the raw JSON
result to be sent back to the client:
```json
{
"commands": [
{
"selector": "div#result",
"action": "replaceInnerHTML",
"params": {
"html": "<p>Parameters: widget2000, firstpage.html<\/p>"
}
}
]
}
```