BuddyPress.org

I said I wasn’t going to document the code yet. I will tell you how it works though.

bpc allows any type of content to be organized into a hierarchy. Think of it as a virtual file system implemented with SQL tables and relationships. Each path in the tree can have it’s own set of tags and content items.

bpc 1.0a1 creates a tree consisting of member, group and blog tags.

root

— user

— group

— blog

That’s the tree for bpc 1.0a1. All content, tags or items, are children of the tag ‘root’. Root is invisible to everyone and it’s the only thing allowed at it’s level. All other content is organized as children of ‘root’.

Below root are the tags ‘user’, ‘group’ and ‘blog’. These tags are invisible to the end users also. When a bp member tags their profile they create other tags that are forced to be children of ‘user’.

root

— user

—- cats

—- dogs

—- fish

These tags form the member tag cloud. That cloud template tag looks in the path ‘root/user’ and displays everything it finds there. I’m just going to stop saying ‘root/user’ and use ‘/user’. The root is implied by the first slash. Just like a file system.

What actually happens further is that an content item representing the user who created the tags ‘cats’, ‘dogs’ and ‘fish’ tags is put in each tag. We wind up with this:

root

— user

—- cats

---

item (burt)

—- dogs

---

item (burt)

—- fish

---

item (burt)

These ‘items’ are content neutral representations of any type of content. In the case of items in /user, all items are derived from the bp user that created the tag. In the case of this tree some people want to be tagged differently and some the same:

root

— user

—- cats

---

item (burt)

---

item (john)

---

item (jeff)

—- dogs

---

item (burt)

---

item (john)

—- fish

---

item (burt)

—- giraffes

---

item (jeff)

—- bears

---

item (john)

When a user creates their own set of individual tags bpc inserts an item for them into an existing tag or creates a new one.

So we’ve got a tag cloud now. In bpc the links in the tag cloud launch the content specific bp directory. A member tag cloud launches the member directory with a query param like /members?tags=cats

The member directory has been modified by extending the normal template class that bp uses to pay attention to those $_GET vars if they exist in the url.

The directory parses the $_GET query vars and determines what tag the user clicked on. In this case the tag was ‘cats’. The directory template class then gets all the items in the path /user/cats. These things it gets back from bpc are not the actual user records we need to display in the directory, they are instances of the class OCI_Item. That neutral representation of the original content.

We get neutral representations of original content from classes that understand the original content types. They know where they came from and how to convert them to OCI_Items. Each content type that wants to store info in bpc has to have one of these conversion classes. In the case of bp members the class is called OCI_Item_User and it is a derived class of OCI_Item.

It’s defined like: OCI_Item_User extends OCI_Item. You’ll see it in the oci-classes.php file. There’s three like that in there. One each for users, groups and blogs. The purpose of these classes is to convert source content into item format and when the time comes, take the item information and get() the original item back.

This is the template class for the member directory:

class OCI_BP_Core_Members_Template extends BP_Core_Members_template.

When this class is fired up by the user clicking on a tag in the member tag cloud, the directory is launched. The template is run and in the constructor for OCI_BP_Core_Members_Template, it does a bunch of stuff and gets to the following chunk of code (lines 170-188 in oci-classes.php):

} else if (isset($_GET[$bp->contents->slug])) {



if (!$path)

return;



$tag = $path . '/' . $_GET[$bp->contents->slug];



$container = new OCI_Container('path=' . $tag);

$items = $container->get_items();

foreach ($items as $item){

$tmp->user_id = $item->item_id;

$users['users'][] = $tmp;

unset($tmp);

}



$users['total'] = count($users['users']);

$this->members = $users;

} else {

If there are any query vars such as /members/?tags=cats then instantiate an OCI_Container for the path ‘/user/cats’ and get all the items in that tag.

When the OCI_Item_User creates each member item in a tag it stores the user id in the var $this->item_id. The group version and the blog versions of that class store the group or blog ids in that spot when they create items.

I took a look at the the bp template class and figured out that it needs an array of the format: array(‘users’ => $users, ‘total’ => $count) to operate properly so that’s what gets built for it in that chunk of code.

Group and Blog templates use a similar thing. So we just pull out the user ids stored in each item and create an array that bp’s template class is happy with.

All I had to do to modify the the directories was create a new bp_has_whatever() tempate tag and override the constructor for each template class that bp uses.

Now we’ve got some modified php templates, classes and a replacement has_whatever() template tag function. It’s a good thing that Andy is filter and action happy because the only other thing to do was to create a filter that tells bp to use our new directories instead of the standard ones.

// changes the default members dir to this one

function oci_filter_template_directory_members(){

return ('oci-contents/directories/members/index');

}

add_filter('bp_core_template_directory_members', oci_filter_template_directory_members');

That function lives in oci-templatetags.php. There’s one for each of the directory types being hijacked.

yes i just see the tag form.

at https://buddypress.org/forums/ very top on the sidebar ,

the report bug button takes me to forbidden. (not 404)

i guess maybe it need a separate account ??

The chat interface has been revamped to better support chat with many different people at the same time.

So far it works in Firefox, IE 7, and Chrome. Not sure about Opera or Safari. IE 6 is in development, although I still say that using IE 6 ought to be illegal….

If you find bugs, or use Opera or Safari and test the chat please report whether there are problems.

http://simplercomputing.net/bp

We’ll be adding audible alerts when a new chat message arrives and some other cool features soon.

Oh. Before anyone publicly complains, no, there isn’t any pagination included in alpha 1’s directories. I’m totally ignoring it on purpose. All the code is there ready to take advantage of it but I purposely didn’t implement it.

bpc uses a mechanism called Modifed Pre-Order Tree Traversal (MPTT) that was specifically designed to track tree like structures in SQL tables. I need to find out if that part of the core is working properly. I had no intention of introducing another complexity down at the SQL level right now. As far as I can tell the MPTT implementation is stable.

I swore I wasn’t gonna document this for programmers yet but I guess it wouldn’t hurt to talk about MPTT now.

When you look at the properties for a node, in the class OCI_Node, you’ll notice two vars $this->lft and $this->rgt. These are MPTT specfic vars that represent the parent/child relationships of all nodes in the tree. . See these for info about MPTT:

http://www.sitepoint.com/article/hierarchical-data-database/

http://dev.mysql.com/tech-resources/articles/hierarchical-data.html

http://www.dbmsmag.com/9604d06.html

A simple node insert operation in the tree has a good chance of changing 50% of the lft/rgt values in the entire tree.

Any sql operation that changes these values must:

1) lock the table

2) get the node and it’s valid lft/rgt values

3) do the operation

4) unlock the tree

The only reliable property for a node in the tree is the $this->id which is the record id in the table.

The boys and girls at wp are investigating MPTT right now for a future version of wp. I can’t wait around until they get to that so I’ve implemented this in bpc. It’s a natural for this kind of project and has many advantages.

Ha! Yep. Andy is right. Delete that line and things should work alot better. Oops. Fixing the download. Thanks Andy.

3000+ lines of code and one var_dump() included for free.

@EvilChicken

Yes, I believe bbPress is already improving quite rapidly; while it has nowhere near as many features as vB, my instinct is that the combo of BuddyPress + bbPress is actually more useful overall, especially when you consider how the forums are spread out among groups, giving a more intuitive and “lighter” feel.

bbPress 1.0 should be released with the next week or so (it is currently on RC1) and it’s development has been given a boost by the resources Automattic are pouring into their new commercial service, TalkPress.com.

Upcoming bbPress improvements include easier integration with WP and MU, which will be a huge relief. If the massive WordPress community starts to swing behind bbPress, we can expect some pretty good plugins to emerge, whereas I believe vB’s community of voluntary plugin developers has already started to hemorrhage. Certainly, the fact that they are closing all threads mentioning BuddyPress suggests they feel threatened by it.

Jeff, thanks for the extensive efforts to repro on your side. I have answers to your first two questions:

1. No other plugins

2. Default BuddyPress Home Theme

I will investigate the server logs.