Code Monkey Ramblings

I'm a guy who used to run MT and WP blogs, on a shared host. I recently decided to get back in the game. So, I grabbed MT 4.25 commercial from ProNet (I'm a member) and installed it at my host. I futzed around for a few days, trying to remember what I'd forgotten about templates and such. Then I remembered that no one is paying me for all this, and downloaded WP.

I'm not qualified to say if MT is technically superior to WP, or vice versa. (Although I do agree that most blog content never changes and, hence, should never get near a relational database. If Dave Winer or someone else would write an updated Radio Userland -- i.e., my files live on my desktop and are pushed to the server -- I'd plug in Disqus and go to town.)

While it's inane to argue that MT should be rewritten in PHP simply because Perl programmers are older, no one other than developers cares about that debate. The thing could be written in Cobol for all the rest of us care.

With that in mind, WP and the community surrounding it has so much going for it. People -- bloggers who don't do it for money -- are attracted by the zillion themes available for WP, by all the plugins, and by all the tech-clueless people just like themselves who populate Wordpress world. It's just damn easier to do the things most of us want to do.

If you are being paid to write code, if you are being paid to stand up a separate server just to handle comments, then MT is probably your best bet. But, if you aren't being paid, if you're just a guy renting space on a server to run a personal blog for kicks, then why mess with all that? Why not install WP?

If SixApart is interested in increasing market share among non-commercial customers -- meaning, non-paying, so why would they? -- the best investment they could make would be to hire some folks to design and build a boatload of distinctive templates and styles, and to conjure up some kind of brain-dead simple installation gizmo.

But, why would any business do all that for people who won't be paying customers?

While it’s inane to argue that MT should be rewritten in PHP simply because Perl programmers are older, no one other than developers cares about that debate. The thing could be written in Cobol for all the rest of us care.

Well, there are good developer reasons why SixApart might want to adopt PHP 6, and one of them that I cited in the forum was that the Perl community at large seems to not give a rodent's posterior about actively maintaining a runtime that can compete with offerings from other languages like PHP on performance and debugging. PHP 5.3 introduced namespaces, among other things, and now the gap between PHP and Perl is closing rapidly from the perspective of many developers.

One of the major problems that Perl has is that mod_perl simply sucks compared to mod_php. Because Perl runs as a CGI binary, not a web server module, it has to be reloaded as a distinct process every single bloody time you make a request. mod_php takes away a lot of that performance issue by embedding the PHP runtime directly into Apache, ngix or whatever server you are using. It's been key to the rise of PHP as a serious competitor to Perl, and again, the Perl community simply doesn't take it very seriously.

PHP 6 will be a very compelling language. It would also offer SixApart a chance to do a radical, "MacOS X-like" rewrite of Movable Type to make it far easier to debug, more attractive to a larger base of developers, and faster because of the gains to be had from mod_php.

Because Perl runs as a CGI binary, not a web server module, it has to be reloaded as a distinct process every single bloody time you make a request. mod_php takes away a lot of that performance issue by embedding the PHP runtime directly into Apache, ngix or whatever server you are using.

See FCGI.

PHP 6 will be a very compelling language. It would also offer SixApart a chance to do a radical, “MacOS X-like” rewrite of Movable Type to make it far easier to debug, more attractive to a larger base of developers, and faster because of the gains to be had from mod_php.

MT running under FCGI is plenty fast enough as it is, but honestly I think you're missing the point of the static publishing vs. dynamic publishing debate entirely. Speed gains to be had -- if any, and I question if there are any to be had moving away from PERL -- is going to be marginal at best because the bulk of MT is in the backend that is only used when publishing or managing.

99.9% of the traffic to a MT site is only going to be hitting static HTML which is insanely fast.

I don't want to take this discussion off-topic into that debate, but I think Mike that you're not quite as up on these things as you think you are.

This post came up on some of the MT development mailing lists, where I posted my thoughts. It has been requested that I post them here as well, so here they are in full. This represents my first reaction and probably the rest of what I have to say on this matter. No offense is intended, just being honest here.

These are my opinions alone, and not those of 6A or anyone else on those mailing lists.

Movable Type used to be the king of blog software, but it has since gotten its ass handed to it in the marketplace of ideas by WordPress.

Ideas? No. Wider appeal to an audience that has been trained since birth to think that they are too stupid to use complex applications? Yes. Microsoft has been handing every other desktop operating system its ass since as long as I can remember, but that doesn't mean that those other systems are doing something wrong, or that Microsoft is doing something right with Windows, or that Windows is better in any imaginable context.

Sometimes the market goes one way even when it's the wrong way, and you've just got to live with it.

WP is the Windows of content management applications and it'll take a sea change moment to knock it off. MySpace isn't coming back against Face Book and Yahoo! isn't coming back against Google.

Yet that doesn’t change the fact the community is just listing in the waters, without a lot of active development from outsiders like WordPress.

I would ask this guy why is it that Movable Type needs an active development community if it's already a superior product, and wouldn't MT being a superior product prove that you don't need an active open source community to be successful?

And that's to say nothing of MT and WP targeting people with different needs.

Designers? Now that is a sad state of affairs. There are only a handful of us that take care of things on that front.

Overrated. I don't know any successful site that uses a stock template, if you're serious about your site -- it is my understanding that MT is targeted to people who are serious about their sites -- then you're going to create your own themes regardless. If people don't know enough to know that visual branding matters, chances are they aren't going to need the things in MT that make MT more useful than WP.

If that is true, then having a plethora of free themes really isn't going to change the underlying nature of that relationship.

When I see five WP sites all using the same theme, I cringe. And if those people are going to be lured from one CMS to another because of themes, then they might as well be using Blogger, because Google offers everything they are likely going to need and use on a daily basis in a blog.

Threads like this on the Movable Type forums are problematic for me. I mean really, why should anyone have to buy a feature to password protect their entries when systems like WordPress provide that as a basic feature?

This is somewhat valid, but I'd just as quickly point out that WP has features like this but has security problems precisely because of these misplaced priorities. If those guys focused on security as much
as they ought to be doing, they might not have a plethora of features to suit every vocal minority of their user base.

Not to mention the fact that as has been discussed here, entry protection is a heck of a lot more involved with static publishing than with dynamic publishing. Which would people rather have, a site
that scales but has slightly more involved processes for securing entries, or a site that has this "basic feature", but completely vanishes off the Internet under a heavy load?

These are either-or scenarios that can't be so quickly introduced and then dismissed. If WP had the scaling ability of MT *and* it had the "basic feature" of entry protection, then that might be a legitimate gripe. But when you've got trade offs -- intentional trade offs -- then you've got the beginnings of a debate (and a debate on architecture, really, not features), but not a hit list.

I'd also argue that such password protection isn't a basic feature. Having each post go into its own page is a basic feature. Being able to edit old posts is a basic feature. Entry protection is advanced no matter how you slice it.

For blogging, I have to say that WordPress is probably more than enough.

I would agree, but only by rephrasing as such: "WordPress is probably more than enough for bloggers that would otherwise be happy on Blogger."

I used Blogger for over a year and it has steadily improved to the point where I'm not trying to be insulting here. AmericaBLOG, one of the biggest political blogs on the Internet, is run off Blogger. But I
really do believe that when people say "WP is probably more than enough [for the vast majority of users]", they don't realize that such a thing isn't a compliment, it's a marketing pitch.

"Probably" being "enough" isn't good enough, nobody should settle for that. And as we've seen with examples all around us (Windows), that doesn't even necessarily mean that product X is even a good product at all.

That’s also caused me to fail to accept the fact that Movable Type really is slow–too slow–for a shared host because it is so much more capable for an enterprise environment as a content management system than WordPress.

I'm confused about this, has this guy tried MT with the new caching features in 4.x? I run MT on Dreamhost with 20-30 other users on my server and I have never had a problem. I don't get a ton of traffic mind you, but since that's more about Apache serving static HTML than anything, it's really just the backend. Republishing my ~900 entry site takes about 2-3 minutes, which is a hell of a lot faster than it used to be.

I think if you add up the fractions of a second that it takes to generate a WP page for every single hit, it'll eventually add up to far more than you'd spend with the MT backend over the course of a
month doing publishing and management. You have to seriously screw with WP with memory caching, caching plugins, and squid to get WP to behave well.

On a shared host, I'd disagree vehemently, I'd say that WP is unsuitable for shared hosting. I've seen people get kicked off shared hosting because of WP, but never because of MT.

Anyway, just my two cents..

Overrated. I don’t know any successful site that uses a stock template, if you’re serious about your site - it is my understanding that MT is targeted to people who are serious about their sites - then you’re going to create your own themes regardless. If people don’t know enough to know that visual branding matters, chances are they aren’t going to need the things in MT that make MT more useful than WP.

If that is true, then having a plethora of free themes really isn’t going to change the underlying nature of that relationship.

When I see five WP sites all using the same theme, I cringe. And if those people are going to be lured from one CMS to another because of themes, then they might as well be using Blogger, because Google offers everything they are likely going to need and use on a daily basis in a blog.

I don't think your understanding is entirely correct. Movable Type is targeted both as a comprehensive CMS and a blogging tool, to both audiences. It may be targeted more toward people who run larger sites and who have the resources needed to make it run properly, but that doesn't mean that SixApart is abandoning the target audience of products like WordPress. For many of these users, the lack of styles and templates is a drag, and I have seen some bloggers with a large volume of visitors who use stock templates, so you ought to not write them off immediately.

Not to mention the fact that as has been discussed here, entry protection is a heck of a lot more involved with static publishing than with dynamic publishing. Which would people rather have, a site that scales but has slightly more involved processes for securing entries, or a site that has this "basic feature", but completely vanishes off the Internet under a heavy load?

Why does it have to be either? With regard to WordPress, WordPress can function very quickly under a heavy load if you have the Super Cache plugin enabled, so WordPress has both the speed and the basic feature there.

As I said, your response is to let the perfect be the enemy of the good. Movable Type is so anal about being non-discriminatory toward all of the different dynamic publishing systems that it doesn't provide even a basic level of support for PHP, which the majority of its users can or do run if they want that feature set. I've seen this sort of thing done with the Privacy plugin, and it was incredible while it lasted.

I’d also argue that such password protection isn’t a basic feature. Having each post go into its own page is a basic feature. Being able to edit old posts is a basic feature. Entry protection is advanced no matter how you slice it.

I was able to implement basic password protection with 2 Custom Fields and a little PHP glue.

I used Blogger for over a year and it has steadily improved to the point where I’m not trying to be insulting here. AmericaBLOG, one of the biggest political blogs on the Internet, is run off Blogger. But I really do believe that when people say "WP is probably more than enough [for the vast majority of users]", they don’t realize that such a thing isn’t a compliment, it’s a marketing pitch.

"Probably" being "enough" isn’t good enough, nobody should settle for that. And as we’ve seen with examples all around us (Windows), that doesn’t even necessarily mean that product X is even a good product at all.

True, to a point. However, you have to be realistic here. For a lot of people, WordPress just works for them. You could also say the same thing about MySQL and even PostgreSQL compared to Oracle or SQL Server, but it would not be realistic to expect someone running a personal blog that gets less than the tens of millions of hits every month to even consider one of those high-end databases that has all of the serious features.

Is a cheap entry level car an outstanding product? Usually not. A $40k-$50k sedan could usually blow it away as a family car, but for many, the former is more than enough to meet their needs. The key here is whether or not the product is good for what it is.

I’m confused about this, has this guy tried MT with the new caching features in 4.x? I run MT on Dreamhost with 20-30 other users on my server and I have never had a problem. I don’t get a ton of traffic mind you, but since that’s more about Apache serving static HTML than anything, it’s really just the backend. Republishing my ~900 entry site takes about 2-3 minutes, which is a hell of a lot faster than it used to be.

Serving up the static content was, and is, fast enough for me. It's publishing comments. I've gotten a number of complaints about that, and I spent a long time optimizing and tweaking my use of modules, publishing queues, etc. to make it as fast as I could possibly make it, and it still takes 5-10 seconds to post a comment these days (my best was 2-3).

I think if you add up the fractions of a second that it takes to generate a WP page for every single hit, it’ll eventually add up to far more than you’d spend with the MT backend over the course of a month doing publishing and management. You have to seriously screw with WP with memory caching, caching plugins, and squid to get WP to behave well.

SuperCache works very well. It was the first plugin I enabled when I started testing out WordPress 2.7.1 here, and unless I get dugg several times a day, I expect it will still be good enough to make this issue irrelevant.

Anyway, just my two cents..

Thanks for your feedback. I'm sure that you were probably more civil than some of the comments :)

My goal is not to insult Movable Type. In fact, I still plan to use it for most of my site because it is far better for broader site management than anything else I have seen. I just realized, after playing around with WordPress 2.7, that WordPress has come a long way, and that it probably meets basic blogging needs as well, if not better. I don't claim to be running a large blog that gets vast volumes of traffic or anything like that.

...has this guy tried MT with the new caching features in 4.x? I run MT on Dreamhost with 20-30 other users on my server and I have never had a problem...

Well, I'm not that guy, but I've no reason to doubt that assertion.

But, in terms of adoption, market share, or whatever we call it, most would-be bloggers will never get that far with MT. To find out that MT runs pretty snappily these days, and that enabling FastCGI can be as simple as adding two lines to a file, people actually have to build an MT site.

Yes, if you've got the right release, the Professional Website Templates look pretty good, and getting them onto your site is dead easy. However, unless people want to run the MT equivalent of Kubrick, redesign is sure to follow, which means a foreboding dip in the MT template pool.

I want to be clear: I'm not dissing MT's template design. I'm saying for people who aren't being paid to do it, MT's templates have such a steep learning curve that most will just walk away.

I'm not convinced that SixApart has any reason to tweak its code in an effort to go after a market that really doesn't exist: Amateurs who almost certainly wouldn't buy from SixApart.

Nor am I convinced that greater adoption of MT among that community will boost SixApart's sales. Yes, you get more MT visibility if more amateurs use it, but how many businesses are gonna buy into an MT deployment simply and only because Cousin Eddie uses it for his high school football blog?

Bottom line: MT scares away amateurs, but there's no money there. Now, if it scares away pros, that's another story.

Serving up the static content was, and is, fast enough for me. It’s publishing comments.

But that was running PERL as regular CGI, where the interpreter has to be launched, the code compiled into byte code, and all the modules loaded, database connections established, etc, for every single request, right?

If you run MT under FCGI it keeps the scripts running in memory and only runs specific sections for each new request. Modules already loaded stay loaded, database connections can be persistent, etc.

I'd say that the overwhelming majority of time spent publishing very small jobs (like publishing a single comment) is being spent on launching the PERL interpreted and spinning it up, which doesn't really apply under FCGI.

And that is a problem for me. I can't get FCGI on my host, and I don't want to switch hosts because they have a very stable environment for hosting WordPress and Movable Type. I've never had a serious problem with them in terms of up time, data loss or missing modules that wasn't an "act of God" sort of thing. I do know where you are coming from here, and as I said, I'm not dissing Movable Type in the least. I am just no longer convinced that it's the right choice for me and others who want to run a simple blog. For anything beyond that, I completely agree with you. It is nothing short of insanity that some of the major news sites like ZDNet and CNet are running WordPress.

You can also try Host4Pro !I use them for Moneybookers Web Hosting (http://www.host4pro.com/payment/moneybookers/) And Unlimited Web Hosting (http://www.host4pro.com/hosting/)