3 files changed, 58 insertions, 17 deletions
diff --git a/doc/plugins/aggregate.mdwn b/doc/plugins/aggregate.mdwn
index 574c8b125..ecca0232e 100644
--- a/doc/plugins/aggregate.mdwn
+++ b/doc/plugins/aggregate.mdwn
@@ -1,10 +1,10 @@
-[[template id=plugin name=aggregate author="[[Joey]]"]]
-[[tag type/useful]]
+[[!template id=plugin name=aggregate author="[[Joey]]"]]
+[[!tag type/useful]]
 
 This plugin allows content from other feeds to be aggregated into the wiki.
-Aggregate a feed as follows
+Aggregate a feed as follows:
 
-	\[[aggregate name="example blog"
+	\[[!aggregate name="example blog" dir="example"
 	feedurl="http://example.com/index.rss"
 	url="http://example.com/" updateinterval="15"]]
 
@@ -13,12 +13,17 @@ more frequently than once every 15 minutes, and puts a page per post under
 the example/ directory in the wiki.
 
 You can then use ikiwiki's [[ikiwiki/blog]] support to create a blog of one or
-more aggregated feeds.
+more aggregated feeds. For example:
+
+	\[[!inline pages="internal(example/*)"]]
 
 ## setup
 
-Make sure that you have the [[html]] plugin enabled, as the created pages are
-in html format. The [[meta]] and [[tag]] plugins are also recommended. The
+New users of aggregate should enable the `aggregateinternal => 1` option in the
+.setup file. If you don't do so, you will need to enable the [[html]] plugin
+as well as aggregate itself, since feed entries will be stored as HTML.
+
+The [[meta]] and [[tag]] plugins are also recommended. The
 [[htmltidy]] plugin is suggested, since feeds can easily contain html
 problems, some of which tidy can fix.
 
@@ -31,7 +36,7 @@ crontab entry:
 Alternatively, you can allow `ikiwiki.cgi` to trigger the aggregation. You
 should only need this if for some reason you cannot use cron, and instead
 want to use a service such as [WebCron](http://webcron.org). To enable
-this, enable on `aggregate_webtrigger` in your setup file. The url to
+this, turn on `aggregate_webtrigger` in your setup file. The url to
 visit is `http://whatever/ikiwiki.cgi?do=aggregate_webtrigger`. Anyone
 can visit the url to trigger an aggregation run, but it will only check
 each feed if its `updateinterval` has passed.
@@ -59,9 +64,44 @@ directive:
 * `tag` - A tag to tag each post from the feed with. A good tag to use is
   the name of the feed. Can be repeated multiple times. The [[tag]] plugin
   must be enabled for this to work.
-* `template` - Template to use for creating the html pages. Defaults to
+* `template` - Template to use for creating the aggregated pages. Defaults to
   aggregatepost.
 
 Note that even if you are using subversion or another revision control
 system, pages created by aggregation will *not* be checked into revision
 control.
+
+## internal pages and `aggregateinternal`
+
+This plugin creates a page for each aggregated item. 
+
+If the `aggregateinternal` option is enabled in the setup file (which is
+recommended), aggregated pages are stored in the source directory with a
+"._aggregated" extension. These pages cannot be edited by web users, and
+do not generate first-class wiki pages. They can still be inlined into a
+blog, but you have to use `internal` in [[PageSpecs|IkiWiki/PageSpec]],
+like `internal(blog/*)`.
+
+For backward compatibility, the default is that these pages have the
+".html" extension, and are first-class wiki pages -- each one generates
+a separate HTML page in the output, and they can even be edited.
+
+That turns out to not be ideal for aggregated content, because publishing
+files for each of those pages is a waste of disk space and CPU, and you
+probably don't want to allow them to be edited. So, there is an alternative
+method that can be used (and is recommended), turned on by the
+`aggregateinternal` option in the setup file.
+
+If you are already using aggregate and want to enable `aggregateinternal`,
+you should follow this process:
+
+1. Update all [[PageSpecs|ikiwiki/PageSpec]] that refer to the aggregated
+   pages -- such as those in inlines. Put "internal()" around globs 
+   in those PageSpecs. For example, if the PageSpec was `foo/*`, it should
+   be changed to `internal(foo/*)`. This has to be done because internal
+   pages are not matched by regular globs.
+2. Turn on `aggregateinternal` in the setup file.
+3. Use [[ikiwiki-transition]] to rename all existing aggregated `.html`
+   files in the srcdir. The command to run is
+   `ikiwiki-transition aggregateinternal $setupfile`,
+4. Refresh the wiki. (`ikiwiki -setup your.setup -refresh`)
diff --git a/doc/plugins/write.mdwn b/doc/plugins/write.mdwn
index 6b49ec58d..4dc55e302 100644
--- a/doc/plugins/write.mdwn
+++ b/doc/plugins/write.mdwn
@@ -412,12 +412,13 @@ Aborts with an error message. If the second parameter is passed, it is a
 function that is called after the error message is printed, to do any final
 cleanup.
 
-Note that while any plugin can use this for a fatal error, plugins should
-try to avoid dying on bad input when building a page, as that will halt
-the entire wiki build and make the wiki unusable. So for example, if a
-[[ikiwiki/PreProcessorDirective]] is passed bad parameters, it's better to
-return an error message, which can appear on the wiki page, rather than
-calling error().
+If called inside a preprocess hook, error() does not abort the entire
+wiki build, but instead replaces the [[ikiwiki/PreProcessorDirective]] with
+a version containing the error message.
+
+In other hooks, error() is a fatal error, so use with care. Try to avoid
+dying on bad input when building a page, as that will halt
+the entire wiki build and make the wiki unusable.
 
 #### `template($;@)`
 
diff --git a/doc/plugins/write/tutorial.mdwn b/doc/plugins/write/tutorial.mdwn
index 8b12fd183..94b72c763 100644
--- a/doc/plugins/write/tutorial.mdwn
+++ b/doc/plugins/write/tutorial.mdwn
@@ -169,7 +169,7 @@ be a guard on how high it will go.
 		}
 		my $num=$last{$page}++;
 		if ($num > 25) {
-			return "[[fib will only calculate the first 25 numbers in the sequence]]";
+			error "can only calculate the first 25 numbers in the sequence";
 		}
 		return fib($num);
 	}
@@ -182,7 +182,7 @@ does for numbers less than 1. Or for any number that's not an integer. In
 either case, it will run forever. Here's one way to fix that:
 
 		if (int($num) != $num || $num < 1) {
-			return "[[fib positive integers only, please]]";
+			error "positive integers only, please";
 		}
 
 As these security problems have demonstrated, even a simple input from the