WP 3.0 recognizes $table_prefix, confirmed by Ron Rennick.

So WordPress has two ways to do multi-site in WP 3?

One-Install Multi-Site with $table_prefix

1. The original way, before WordPress MU. Requires no plugin. Works out-of-the-box, just change $table_prefix for each new blog in wp-config.php. I believe the instructions in wp-config.php are self-explanatory, but there’s more information in the codex if you need it. This is what I use here with 9 blogs. This is what I recommend if you want a small blog network without too much complexity.

People will tell you this way requires multiple WordPress installations but that’s not true. You can configure your web server to use the same directory for all of your domains. One install. So if you change your theme, as you normally would with one blog, your changes are automatically visible on all of your blogs! If you want to see my Nginx config, ask me–I will show you how to do this with Nginx.

By the way, Nginx is faster than Apache. It takes less than 1/100th of a second for Nginx to serve this blog. Do the math. I’m sure it’s possible to do one-install multi-site with Apache too, but if you look at the growth of Nginx it’s clear to me the future is Nginx. So why bother with (slower) Apache?

Nginx, the little Russian web server taking on the giants | Royal Pingdom

If you look at growth in terms of percent, nginx has been crushing every other web server out there. In terms of absolute number of websites, nginx added 12.9 million websites in 2009. It even kept pace with Apache’s overall growth for most of the year, some months even surpassing it. As a comparison, Apache grew by 13.3 million websites in 2009.

Also, I have seen tutorials to do Multisite with symlinks. This is possibly an option if you don’t have access to your webserver config, for whatever reason.

But think about it, VPS hosting is affordable now. VPS hosting not only gives you access to your webserver config (to make your blogs faster than ever) but you also get access to your PHP config and MySQL config!

The Residue of MU

2. The “MU” way. The official name is “Multisite” now and an entire forum is devoted to it.

What was MU? I believe it was created to build big blog networks of user-generated content–similar to Tripod, Geocities, Myspace, etc.

A terminology nightmare: blogs, sites, networks, and the super admin | Andrew Nacin

MU, or multi-user, was a fork of the codebase designed for many blogs and many users, all on one install.

Here’s a list (Thanks to Mike Schinkel) of schools using Multisite:

http://blogs.law.harvard.edu/
http://umaine.edu/
http://blog.usf.edu/
http://blogs.mcgill.ca/
https://blogs.bgsu.edu/
http://ucalgaryblogs.ca/
http://umwblogs.org/
http://blogs.ubc.ca/
http://www.apu.edu/blog/
http://blogs.fu-berlin.de/
http://blogs.unimelb.edu.au/

I only had time to look at the Harvard link. They keep the riffraff out by requiring a Harvard email address. Hey Harvard, hook me up with a blog ;-)

I admit there are some good arguments for building big user-generated blog networks. Brad Feld even suggests people are more motivated to produce when there’s no payment for the work involved. It’s a complex subject and I don’t want to get too sidetracked.

I mostly passed on WordPress MU because it’s extra code I don’t need. Generally speaking, I believe less code is better.

So MU is now merged into the core of WordPress. However, as far as I can tell, Multisite is entirely optional and not turned on by default. Looking forward, I don’t think we know yet how Multisite incorporates networks of blogs created using $table_prefix, without MU.

The Future

I intend to keep the “$table_prefix method” alive as long as possible, as long as it makes sense to.

The $table_prefix method does have weak points. Using $table_prefix for multiple blogs you have two options tables! In a way, this is a good thing. On newer blogs I’m a little more attentive–I want email notifications when I get a comment. But on my most popular blog I don’t want email notifications, because there is no way I could keep up with the crazy questions. On the other hand, you probably don’t want to maintain multiple moderation/spam keyword lists.

Update: I discovered Multisite has similar options-related weaknesses. So there’s no escaping it.

A terminology nightmare: blogs, sites, networks, and the super admin | Andrew Nacin

A lot of discussion centered around the options API. There are no options across an entire MU multi-domain install, but we still thought about that and it caused plenty of confusion. We also considered get_sitewide_option(), get_domain_option(), get_multisite_option(), and yes, even get_thingieswide_option(). In despair, one suggestion was thing, thingy, and thingies, to replace blog, site, and network.

As far as options management for my $table_prefix network, I wrote a plugin to help maintain multiple options tables. It’s really straightforward and there’s no “super” admin, subdomain admins, etc. I call it oSync. It essentially duplicates options from a “master” blog you choose–probably the one you use most often. Let’s say Domain1.com is the master and you’re in Domain2.com’s admin panel–oSync compares Domain2’s options with the master and shows you the difference. One click changes each option. It actually says something like, “This option is now X, but could be Y.” It needs a little more work before I can release it. If you want this plugin, let me know.

I’m thinking about upgrading 9 blogs and a couple of plugins I wrote. They all depend on $table_prefix, which you all know is in wp-config.php since before Matt was Ma.tt.

Yes, I know WP 3.0 is multi-site. I’ve been using the recommended $table_prefix for several years now. Yes, it’s time to upgrade. Which is why I’m here. So what is the recommended way now to get all my data working with WP 3.0? I have millions of records stored in MySQL using $table_prefix.

Instead of using 9 wp-config files (doh!) I came up with this really simple way of using the same wp-config.php for all of my blogs:

$table_prefix = substr(str_replace(“-”, ”, $_SERVER['HTTP_HOST']), 0, 4) . ‘_’;

The dash deal was added later for my “wp-fun.co.uk”

In plain English, this code simply uses the first four letters of the domain name for the $table_prefix. So for example, Knowingart.com would be “know_” instead of “wp_”. I use this same technique for all my plugins.

So a Knowingart.com plugin would use “know_myplugin” to store information, etc.

What do you think? Can I safely upgrade to WP 3.0?

—Update—

Mike Schinkel from LinkedIn says:

Multisite is a plugin you have to enable. The database layouts should pretty much stay the same as you had before.
…
I’m pretty sure regular sites will upgrade as non-multisite and WPMU sites will upgrade as multisite, and you’ll get the choice during new installations.

That’s my guess too.

It seems MU is the mistress who moved in with the wife, $table_prefix, yet each pretends the other doesn’t exist. They seem to have opposing interests (user-generated content vs. edited content) yet both bring (multi-site) children.

Personally, I’d like to see MU swept under the rug. But I don’t think that’s going to happen. User-generated content brings in lots of cash. Edited content costs more to produce.

How much is WordPress plugin development worth, per hour?

Have you ever paid for anything WordPress-related besides installation/migration?

Do you think WP theme development is worth more than WP plugin development?

Let’s say you have an idea for a WordPress plugin. How much would you pay to have it made for you? $5 $500 $5000 $50,000?

Let’s say you want this plugin for free.

Are you going to write a blog review of it?
Are you going to post a Twitter update about it?
Are you going to keep the backlink ad in the plugin?
Would you pay something to have the plugin customized for your needs?

How else can a developer make money giving away plugins for free?

Maybe I found a bug in WP? You would think a separator of ” would mean no separator. But that’s not the case, WordPress adds space to your title with a ” separator.

The “right” parameter doesn’t work either, that simply moves the space to the other side.

Probably not a big deal, but in terms of SEO: Why stuff your title with non-essential information?

The ‘0′ allows you to trim the title. Otherwise WordPress would echo the title for you, before you could trim it.

<title><?php echo trim(wp_title(”,0)); ?></title>

(Edit this in header.php)

…Continued from Frameworks vs. Hard-Coding Debate

@PJ
ermm wordpress is oop, where’s your argument?

@Jarret There’s no OOP in wp-content, which is where I spend the majority of my time. I rarely change the core code, which yes has changed a lot since I started using it.

You’re right, WP is no longer a simple little script and I’m not spending as much time on it for that reason. The build-measure-learn loop of WP is now too inflexible for significant innovation. So I’m spending more time on my own apps.

With every new version of WP there’s more and more bloat. Take for instance the “widgets” of WordPress. That idea is to make things drag and drop in a standard way for the average person off the street who now wants a blog and is already familiar with how Windows and/or Mac operating systems work. So here we go again, put everything in a box and make us jump through hoops to get there–like I’m back in the 80s when Windows 3.1 came out.

I think Matt Mullenweg said his goal was to make blogging simple enough for his mom. This explains the increased emphasis on Wordpress.com and the hand-me-down direction of Wordpress.org. I’m guessing his “make it easy for mom” realization came later, after WP came included with cpanel and jumped in popularity with webmasters. The problem here is, to make mom comfortable you end up with 10x as much code that does essentially the same thing as before. Yes, now your mom can use it but also it’s harder to customize, and it’s probably slower and eats up more memory.

WP is increasing in popularity but I think its success is somewhat tied to RSS. If Google ever announced they will pay less attention to RSS going forward, I think you would see WP lose favor from that point onward.

The OOP in WordPress looks more like spaghetti to me, it takes a while to find things now. I want to change something but I can’t find the function I need. And now with their crazy caching mechanisms (which seem to change every year) I’m even more lost in the core code. I see functions calling functions calling cache functions calling functions calling functions all over the place (multiple files) that don’t seem to add any value with each additional layer increasingly disconnected from wp-content. You got to draw the line somewhere, how many functions do we really need to post a blog to the Internet?

Is the medium the message too? Do programmers have a resonating influence emanating from deep within the core of web apps everywhere? I say yes. And here there is a great confluence, on the open web. Choose your frameworks wisely. LOL.

Interesting debate tonight in LinkedIn’s “LinkedPHPers” group. I’m not responding directly to Jay Alter, but this “ramp-up to WordPress development” he mentions got my attention:

I’ve actually seen and experienced similar “confusions” where potential employer / clients are concerned who not only don’t understand the difference between a professional PHP programmer and one who has “experience with a framework”, but also ones who truly believe a seasoned PHP programmer doesn’t fit the bill when it comes to programming for WordPress, Druple, Joomla, Typo3, etc. Granted each of those CMS have their own API and / or function set, but the fact remains that aside from the ramp-up required to become acquainted with the function set or class lib, a senior PHP programmer with 10+ years experience, IMHO, has much more of the edge I’m looking for as an employer than any “WordPress” or “Joomla” expert regardless of the length of time they’ve been doing it.

Here’s my response to the WordPress aspect of the discussion:

—

As far as WordPress development, the “ramp up is easy” attitude doesn’t make much sense to me.

I’ve been hacking WP for years. It feels natural for me. Why? WordPress is very similar to the BBS scripts of the 80s and 90s. Instead of Pascal I’m using PHP, instead of ANSI control codes I’m using CSS codes. It’s really the same stuff with faster connections and more people.

WordPress is the natural progression of the dialup BBS, which similarly ran on back-end and front-end scripts. Batch scripts, Pascal scripts, lots of scripting languages you could learn in a day by reading a “read.me” file. I loved it! Because most of the apps back then were not using MS Windows libraries–just about every app had a unique look and feel.

I see 1994 as a major turning point for software development. From my perspective, the trend was to OOP and MS Windows. All this happened around 1994. Bloated and boring was the new normal. I read a lot of books about OOP and Java in 1994. It wasn’t for me–give me a quick and dirty script.

So you can imagine, when I discovered HTML and Perl and PHP and even Javascript, I was happy. With the web you could see content, community, creativity come back into focus. And this wasn’t “Over-Organized Programming” that required (expensive) 10lb. reference manuals. Of course we do need OOP and monster reference books to build the really big, really important apps of the world, in all their glorious complexity and bureaucracy–I don’t deny that. However, these are two different worlds with mostly different concerns.

This is crazy.  WP was already multi-site for years now.  Anyone who wanted multiple sites had multiple years to figure it out.

That’s what $table_prefix is for.

http://codex.wordpress.org/Installing_Multiple_Blogs

Also, this assumes everyone uses Apache–no longer the case.  Lots of people use Nginx now.

Well, at least it’s an option, not turned on by default. I’m sure someone will use it, but not me!

(This was originally posted at Ferodynamics.com, but I’m reposting it here because it relates to WordPress.)

After a few days of research, I decided to add Google Friend Connect to my Ferodynamics.com WordPress blog. (You can’t add all your blogs at once, you must add each individually–this is why I don’t have a GFC gadget here on WP-Fun.co.uk, at least not yet.)

I think GFC is a great idea–what took me so long to discover it?

Now maybe you can see other people who visit here and get a chance to see what they’re interested in. When you click my Google Friend Connect icon you can see other blogs I “joined” using Google Friend Connect.

There are a lot of possibilities. If you have a website, I recommend you give it a try–all you need to do is add the Google Friend Connect widget code to your pages (takes 1 minute) and *bam* you’re hooked into this viral GFC community that spans the entire Internet.

I’m enthusiastic and plan to promote this more. I have heard some bloggers say they don’t “get” Google Friend Connect but I think Google Friend Connect “gets” bloggers. It was created specifically *for* independent bloggers.

Right now, independent bloggers don’t have much in the way of “glue” to tie blogs together. MyBlogLog was promising in that respect before Yahoo killed it, as they tend to do.

I had always hoped the “Gravatar” concept would grow into something like Google Friend Connect, but that never happened. I don’t really blame Automattic, anyone (myself included) could have developed a plugin to push Gravatars further, but there’s only so much time in the day.

That said, I can still imagine some kind of “Google Friend Connect meets Gravatars plugin” for WordPress–I think it would work with enough blogger interest. Google’s approach works, but I would actually like to see it implemented without the popups and AJAX. So the friend matrix hover content (name, description, url) could be indexed like the rest of the page.

For example, I think the title tag GFC uses here falls short:

<img style=”cursor: pointer;“ jsvalues=”.src:getThumbnail($this,32);$h:handle(this,’viewProfile’);title:getDisplayName()“ jstcache=”30“ src=”http://www.google.com/friendconnect/profile/picture/32/B1E2JvSiF0XuQbA8a0W9dJ93R5Ko-f99k4_UXJlfnCF4a8RjW0lXNAQjhYjIhAScYc_uWUuXn1kC-14oMGMJilgmlTaSe95kC_GAg-J8IUy8amokPsNSqolmrNXzQ0yAoSkOQO_y3hg“ title=”PJ Brunet“/>

Come on Google, that’s not really crawlable. If this was a WordPress plugin, the HTML output could read like everything else on the page. Sure, Google has all the data warehoused somewhere, maybe for search rank signals, but I still think it’s not an elegant solution. Maybe a WP plugin could pull in GFC member data via GFC’s API and output more palatable HTML, but a) I haven’t studied GFC’s API yet and b) it’s not a top priority for me right now.

For now, I’m going to sit back and see how many people join my site via GFC, as is. Let’s see how fast this network grows and see what people do with it.

—

Posted to Twitter: I think Google Friend Connect “gets” bloggers http://bit.ly/a5Yufx #googlefriendconnect #blogging #wordpress #foaf #gravatar #mybloglog

Why Diaspora Wont Work

The problem with this “seed” idea: Where will the average person host it? People use Myspace and Facebook because they’re not interested in the complexities of web hosting and/or they can’t afford it.

What they’re describing is a P2P system. The problem with P2P networks without a hub: search is extremely slow, expensive (bandwidth) and not really accurate because you’re routing queries through lots of seed servers.

The only way around it (as we saw with P2P sharing networks) is to have volunteers running the hubs. But I thought Diaspora was trying to avoid centralized hubs?

To make a P2P network searchable (to find people) the seed servers need to communicate with each other–that’s bandwidth the average person won’t pay for.

For comparison, look at something as simple as the Trackback, a very simple way for blogs to automatically link to each other–this is exactly what a “seed” server would need to communicate with other seed servers.

But guess what, nobody uses Trackbacks these days. Why? Trackback spam. So the “seeds” would get lots of annoying “friend” requests from all over the web. So now you have the problem of blacklisting abusers and you’re going to start to feel like a gatekeeper, like Mark Zuckerberg.

Doesn’t Diaspora need a community that already has web hosting and a database? What community already has hosting and a database? WordPress bloggers. WordPress bloggers could help Diaspora get the idea off the ground, but I don’t think they would care.

WordPress bloggers already know how to link to each other. If they’re not already linked to each other, they don’t see any reason to. They’re probably already using Twitter, Skype, Irc, MyBlogLog, Google Friend Connect, forums, or some other app. What does Diaspora have that we don’t have already?

Oh wait, it doesn’t exist yet, how can I compare features?! The fact this non-existent app has $100,000 worth of donations blows my mind.

Good question in my comments:

Why You Should Avoid WordPress

Richard McIntyre

I am wondering why your comparison is with jQuery? JQuery is no good what so-ever to be the main data driver for a website. Even though it is capable (I love and use jQuary myself), but to get some sort of search engine recognized content with almost static pages then any CMS that you implement good caching on will be the tool for a job.

jQuery is a tool to make websites more interactive, not as a base to build interactive websites.

Here’s why.

How many business can really afford the time and money required to maintain “data-driven” websites? In my experience, most people who want a website are on a limited budget and basically want a “brochure” that provides contact info, description of services, etc.

How do you survive as a web designer? You need lots of accounts bringing in monthly hosting fees. So to make a decent living you’re going to end up hosting 100s of websites, if you want to eat. I doubt there are many web designers that have the time/expertise to maintain 100s of WordPress websites–they would have to raise prices and would lose a lot of business, from my point of view.

As it is, I think web design is a horrible business because you compete with every student in the world who found a copy of GIMP (or bootleg Photoshop) and has a $5/month hosting account. I sound pessimistic because I experienced firsthand “unpaid web design internships” in college, and saw how popular “unpaid web design contests” were, and this was way back in the 90s when the tools of the trade were still somewhat new. I’m sure it’s worse now with more hungry, experienced designers living in countries with low living expenses. And now I’m sure Facebook hurts the bottom line of web designers too.

So my argument is WordPress is not a good choice unless the client explicitly tells you they want to blog on a regular basis (most businesses don’t have time to blog) and they want the best SEO possible. Without regular blogging from the business you end up with a dead feed that hasn’t been updated in years.

In reality, if you’re a business professional, I don’t believe your SEO rank is so much determined by the content of your website. It depends more on how many relevant professional organizations point to your page, the age of your domain, the proper credentials/endorsements/licenses, your geographic location according to Google, and your listing in directories like Superpages.