My friend Antone Roundy talks about what’s right … and wrong … with Aweber in this post.

Let me add one to it: Aweber needs an API.
(an API – or application programmiong interface – is a way for programmers to write software that talks to each other.)

Not only don’t they have one, but they shut people down for “scripting” the form.

Yes, they need the users IP address, but they can get that when they click on the confirm link … like the rest of their competitors do.

Most other autoresponders out there make integration easy with simple API capability: GetResponse, iContact, even MailChimp.

Their alternative “email parser” works – usually. But it’s a pain to setup and seems flaky, not always processing the messages sent to it.
A simple API to add subscribers would sure make things a LOT easier to integrate.

That said, I do have a whole series of Aweber based solutions – mostly for Joomla, but some training videos as well – on the Intellispire site.

Yesterday, Amazon released their CloudFront Content Delivery Network (CDN), which finally makes true, fast CDN’s available at an affordable price to mere mortals.

As usual, I jumped on the challenge and quickly implemented a Joomla! 1.5 native plugin for the service. This plugin automatically moves your sites static content (images, css, javascript) to the CloudFront CDN, thereby providing faster loadtimes for your users … especially those users that are geographically “far away” from your servers.

Setting up the service was extremely easy. First, I had to get the content ready and upload it to an S3 bucket. While arguably easy using the s3sync command line tools, the process is complicated by the fact you need to choose exactly “which” files to upload, setting the right headers, and, for more advanced users, gzipping them in advance (more on that later).

Once I had my staging area ready, I just uploaded everything to S3. Then, using the simple but effective PHP based client, I registered the bucket, and received a cloudfront.net domain that can be used interchangeably with the s3 bucket, but is actually on the CDN.

Configuring Joomla! was even easier. I uploaded the CloudFront plugin, told it the current domain and path of the site, and the new path using the CDN. Then I published the plugin – and bingo! All static content is now being served from CloudFront instead of my local server.

A bug or a feature – I haven’t decided which, yet, – is that the not only the front end Joomla! files are being served from the CDN, but the backend files (administrator) are, as well. I may make that ‘configurable’. While I like saving a second or two from each request, I’m not sure I want that to happen in all cases.

A side advantage of this setup – now that my webserver no longer serves static content – is that I can optimize it for PHP.

Before turning on the CDN, I took screenshot of the site (a stock Joomla! site) with YSlow. You can see the image here (big picture opens in a new window):
http://itest.s3.amazonaws.com/nicktemple/images/J-NOCDN.png

The biggest thing to notice is that YSlow gives this site a big ‘F’ for performance … 46 out of a posible 100. But remember, an F is still 1/2 an ‘A’ :->

Now, after turning on the plugin, we do a little better … a ‘C’ grade, with a numeric score of 76.
http://itest.s3.amazonaws.com/nicktemple/images/J-CloudFront.png

Wow! A big improvement!

Note, though that, YSlow doesn’t yet know that ‘cloudfront.net’ is a CDN, so you must tell it, using the instructions found here: http://developer.yahoo.com/yslow/faq.html#faq_cdn

Overall, I’m extremely impressed with the service, and am using for my sites and my largest customer sites. There are two downsides right now:

a) CloudFront doesn’t currently support ‘gzipped’ content, nor does it support ssl. There are some workarounds discussed on the forum, so it isn’t a performance show stopper. It just makes setting up the staging server a little more difficult.

b) Setting up the staging server isn’t particularly “easy”. This is not fault of CloudFront, but rather the fact the tools there aren’t quite ready for most non-developers to use, yet. That’s the portion we’ll be working on over the next few weeks by building a service that allows you to easily manage CloudFront for your Joomla! powered site.

Finally, there are additional performance improvements that can be made to Joomla! For example, combining and minimizing the Javascript and possibly some of the css, as well as caching some of the work the CloudFront Plugin is doing could increase response times for most Joomla! powered websites dramatically.

Signup now (upper right) to be notified when these products are available.

While I’m not yet releasing the plugin except to current customers, if you’d like to hire my company Intellispire to implement CloudFront for your website, please click here to contact us.

P.S. If you are a WordPress or Drupal developer, contact me … I’d like to discuss porting some of our products to other CMS’s, too. Contact me!

I’m taking down an EC2 instance I’ve been running nearly 5 months …

23:15:47 up 149 days, 13:02, 2 users, load average: 0.00, 0.00, 0.00

Up so long that it doesn’t support the persistent storage.

It took me about 2 hours to backup, bring up a new instance, and rsync things over.

Then I moved the ip (it was already using the elastic IP’s), and it’s running fine.

Adding a new 100GB disk was easy, and my old daily “backup to s3″ scripts can be changed to simply take a snapshot. Life has just gotten much simpler.

So now I need to “terminate” the old instance … kind of a harsh word for something that I’ve used daily. By-by i-6279de0b.

I’ve been playing with Jumpbox – very cool technology. Just download the “jumpbox” (the free one is available for Joomla! 1.5 – which is great). This is a copy of a complete web-app in a self contained virtual appliance – both of the virtualization technologies I use – VMware and Xen – are supported, as well as a few others. I’ve been playing with the VMware version.

I did have a minor problem with Firefox 3.0 being picky about the self signed cert on port 3000 (used for initial configuration) – no big deal, but something that could throw off a newbie.

After installing, I immediately installed my Intellispire Add Software component:
http://www.intellispire.com/web/Joomla/download.html

Fifteen minutes later I had a fully configured Joomla install with Forum, Community, decent template, and all the required components to have a real site.

This is the best way I have found to try out new components, do testing, and just play with Joomla! without having to have a server. And with the Updater, I can actually demo building out a site on real-time. Then it’s a simple backup / restore to deploy to a live site …

Can’t get much easier than this! Check them out:
JumpBox

The Aweber Plugin for Joomla! was just released to the Extensions directory this afternoon, and already people are installing it on their sites.

You can visit the Joomla! Extension page directly:

This is a “plugin” that sends a “subscribe” email to Aweber every time someone creates an account on your site, whether through the standard signup process, Community Builder or via a Virtue Mart purchase or AEC subscription.

It can be purchased directly through the site or through the updater.

One question I frequently get is: I’m already using “1ShoppingCart:” (insert name of 1ShoppingCart private label), why do I need Aweber too? They already provide

The answer is actually pretty simple. Aweber does one thing, and they do that one thing well: they deliver email. 1SC and other systems do mail as an “extra feature”, and don’t always get it right.

The bottom line is that if you are serious about getting traditional email delivered, Aweber (and its brands such as JVListPro) is the way to go – even if you are using other solutions that have “build in” email systems as well.

And now, with the Aweber Plugin for Joomla! you get to harness that power with your Joomla! site. Of course the plugin works perfectly well with JVListPro, too.

Matt’s one of the founders of DeanAlert, where I’ve been helping to build out an extremely scalable messaging service based on cloud technology. Matt talks about the past, the future and a little about the memorial day fiasco.

It’s a short 8 minute video …

http://www.kentuckystartups.com/2008/07/14/profile-pearlabs-and-dean-alert/

As I mentioned in my last post, “tags” in subversion is a hack. A copy of a directory sitting in /tags that is still writable is simply NOT what I need when I need to “tag” (remember) a revision (a release, merge point, or a vendor drop for example).

Playing with git, the two things I liked the most about the day-to-day development features were proper branching and merging, as well as tagging. Subversion, in the latest release, is getting near to having proper merge support.

But it still lacks “tags”, “labels” or whatever you want to call them. The most descriptive name would be simply “named revision”. Here’s a simple (perhaps naive) implementation idea.

Assume a tag is simply a named revision. Since revisions are global, so are tag names (vs. being tied to a specific directory … which could be implemented but adds additional, unneeded complexity to the implementation).

This would seem trivial to implement, and could probably be done with a “wrapper script”, that simply does a search and replace on the subversion command before sending it to subversion. I haven’t looked at hacking subversion itself ….

assume::
tags.txt could contain:

release-1.0.0: 29
release-1.1.2: 33
release-stable: 33

Then whenever a command takes a revision, just search and replace:

I type: svn co http://svn.intelliforge.net/svn// @release-1.0.0

subversion sees: svn co http://svn.intelliforge.net/svn// @29

Similarly, you don’t manually “remember” revisions for creating patches, merges or whatever:

I write: svn diff -r release1.0.0:release-1.1.2
subversion sees: svn diff -r 29:33

I’d propose one new command (or a seperate tool). In this trivial implementation, the tag command simply manipulates the tag.txt file, and checks it back in.

svn tag [-f] [-r] [-d] [name]

svn tag simply lists the tag names and associated revisions (svn cat /tags.txt)
svn tag tagname associates the current revision with a tagname, where “-r” specifies a specific the revision,

“force” would overwrite / move a tag name, and “-d” deletes the tag if you want.
You could even specify -m for the commit message, if you wanted to keep “tags.txt” logfile current.

If a tag name exists, an add would fail with a warning unless -f is passed – in which case the old tag is deleted and a new one created (basically a “move”). Note that since tags.txt is, itself, versioned, you could conceivably get a list of all tags over time …

For a native implementation, you’d just have to patch the svn libs so that -r can take a numeric revision or a text label, and do the appropriate lookup.

I’m sure the people who live and breathe subversion can find an appropriate way to implement this or something similar.

Any takers?

I hadn’t heard the term before … SPIT is to telephone what SPAM is to Internet Mail. Having spent the last few months building out the voice technology for Boom Engine, it’s an issue to be aware of.

http://www.freeswitch.org/node/125 | http://it.slashdot.org/article.pl?sid=08/06/12/136232

Coming from an original assembly language (computer machine language) background – where optimization is the norm, I foundChris Vincent’s new site at PHPBench about PHP micro-benchmarking intriguing- it’s well worth the read. You can find a short description here.

While I agree wholeheartedly in this explanatory comment at the end of the page:

All in all micro-benchmarking was not considered to me as something that Is utterly important for how a program works. I agree with those people who have said that the major items at the hardware and algorithms.

I also agree with Chris’s approach: if you can determine the faster way to do things, and train yourself to use the “better” constructs consistently [never at the expense of maintainability, however] you’ll get a step up in your code … not only in performance (which can be significant in large programs), but also in terms of consistency in your code.

I do have some questions about this benchmark, for example, exactly what version of PHP is being used, and what is the platform? Would a 64 bit processor have different charcteristics than a 32bit? Maybe. Maybe not.