Closed GoogleCodeExporter closed 8 years ago
When I say generating documentation for test.js doesn't work, I mean that files
are
generated in the "out" directory, and the "File Index" lists test.js, but I
can't find
any documentation for foo.convert on the "Built-In Namespace _global_" page or
anywhere
else. (I suppose I would have expected a "foo" namespace page to be generated.)
Original comment by ithinkihaveacat
on 4 Dec 2009 at 10:44
I'm sorry you are having trouble getting started. I'm writing a fairly complete
"Getting Started" tutorial right
now actually, but if you take a look at the output from running your example
code above you should see that
JsDoc Toolkit is trying to tell you what the problem is:
>> WARNING: Trying to document convert as a member of undocumented symbol foo.
This means you are adding documentation for the method named 'convert' before
you've documented it's
parent 'foo'. It's easy to fix, just document 'foo'. The following should
produce no warnings:
/** @namespace */
var foo = { };
/**
* Converts a wizzle to a wozzle.
*
* @example
* foo.convert("ffff");
*
* NOTE: Only works if the wizzle is less than 100 characters long.
*
* @param {String} s string to convert
* @return {String}
*/
foo.convert = function(s) {
return "jjj";
};
Original comment by micmath
on 5 Dec 2009 at 10:39
Please consider making this an error instead of a warning, since otherwise it
seems a
bit odd that otherwise valid documentation is tossed away. My reaction to the
warning
was "Yes, I know I didn't document that, I'll get around to it later."
(This is slightly a side effect of me using closure and I'm not actually sure
*how* to
document it, since the namespace is implied by goog.provide / goog.require.
I'm trying
to avoid pulling in the closure sources into my docs, but nothing else
describes the
namespace)
Original comment by jeffbailey@google.com
on 2 Jan 2010 at 8:46
I'm not certain this is an error. In fact the documentation is not "tossed
away" but you are setting up a
situation where you cannot ever get to the documentation.
If you think of the set of symbols in your documentation as a tree, with the
_global_ "class" at the base, the
expectation is that every symbol you document will be on a branch that
eventually connects back to _global_.
However if you start documenting a method "foo" on class "non.Existent," when
there is no documentation for
"non.Existent" then there is no way to get to the branch that foo is on. The
documentation still exists however.
You can prove so by doing the following (pay attention to the "vampire"
method)...
----
// with thanks to www.daveoncode.com for this example
goog.provide("com.jsmonstersbattle.monsters.Monster");
/**
* A common Monster
* @constructor
* @param {String} name
* @param {String} age
* @param {Number} level
* @borrows com.jsmonstersbattle.weapons.RayGun.sap as this.vampire
*/
com.jsmonstersbattle.monsters.Monster = function(name, age, level) {
}
/**
* Returns an amount of damage points
* @return {Number}
*/
com.jsmonstersbattle.monsters.Monster.prototype.attack = function() {
}
goog.provide("com.jsmonstersbattle.weapons.RayGun");
/**
* Reduce a target's level.
* @function
* @param target
*/
com.jsmonstersbattle.weapons.RayGun.sap = function(target) {
}
----
Here we see that the documentation for com.jsmonstersbattle.weapons.RayGun.sap
exists: it must because it
can successfully be borrowed by com.jsmonstersbattle.monsters.Monster. The only
problem is there won't be
any link to com.jsmonstersbattle.weapons.RayGun in the class listing, because
it is never documented as
being a class. For all JsDoc Toolkit knows RayGun might be a String Object, or
a RegExp Object, it won't just
arbitrarily decide that RayGun will be documented as a constructor function or
a namespace object, even
though it appears to you and I that it *probably* is one of those things.
So things will still work with the warning, nothing is being discarded or
broken, you're just doing something
that's quite suspect, and is *probably* not what you meant. I'm leaning towards
leaving this as a warning.
However, if you propose to the user list changing this to an error, I might be
swayed by the community.
Original comment by micmath
on 2 Jan 2010 at 12:57
Original issue reported on code.google.com by
ithinkihaveacat
on 4 Dec 2009 at 10:38