I decided to kill the Coderlog early in its life because it was trying to solve the same problem as coder.io itself. The Coderlog was a way to keep up to date with interesting new things in the world of programming and computer science but.. that's coder.io's entire mission. It seems crazy, then, for me to spend time manually curating links here when I should be improving coder.io to do the same job semi-automatically.

I want to integrate some more human-driven curation options into coder.io itself, but in terms of running a separate blog just about programming stuff, it makes no sense. Sign up at coder.io and subscribe to the tags and queries that interest you instead.

Since coder.io had its soft-ultra beta-"not telling many people yet" type launch less than a month ago, 403 users have somehow discovered the site and decided to sign up - thanks!

The user numbers will explode soon because I have a mailing list of 1700 interested developers I still haven't mailed yet. Oops.

It's of importance I keep pushing forward with new developments each and every week, so here's a public To Do list to show you what I'm working on and to act as a form of social pressure. Feel free to post a comment with ideas of your own.

• Write a series of blog posts about "how coder.io works" covering how content is found, how the system is structured, what technologies it uses, etc.
• Starring and saving of items - for personal recollection (this is a biggie..)
• API documentation
• OpenID support for logins and signups
• Twitter OAuth support for logins and signups
• Add UI for people to submit "leads" to the coder.io system/add items
• Add UI for people to submit sources coder.io should crawl
• Multiviews - view types that can be chosen by users/readers at will (e.g. list, "newspaper style", links only)
• "Official" links against tags so that key resources can be prominently featured
• Avatars and metadata for tags (such as descriptions - helps people choose tags better)
• Advanced query/subscription builder
• Stack Overflow integration
• GitHub integration
• Notices/news box so key improvements can be easier seen
• Remove the box nagging people to leave a message
• E-mail subscription support - so you can have a weekly/monthly/whatever e-mail with content from your subscriptions
• Image extraction from pages - to show relevant pics on items instead of just avatars Canceling this idea, it's frivolous and not necessary for the sort of content coder.io covers.
• Add many more sources (a permanent item)
• Add more tags and refine those existing (a permanent item)
• I will extend this list over time..

Hacker Monthly is a new magazine in PDF and print formats that's based around content that makes it onto Hacker News, a popular social news site for developers and startups.

After seeing the quality of issue 1, I decided to get in touch with them about advertising in issue 2. They cut me a good "first time" deal - it's basically half the published rate - so I thought it'd be silly not to both support them and do a very passive bit of promotion to HN (which, considering coder.io is still in beta, I have not pushed yet).

I'll be doing an analysis post of how the ad performs in a week or two, because I know a lot of you will dig that, but for now, go check out issue 2 of Hacker Monthly. The PDF version is FREE but you can pay to get the print version if you like.

Ian Ozsvald has put together a great blog post with videos and source code demonstrating how he built a face-tracking robot with Python and some electronics parts in a single afternoon. The stages are laid out so you can give it a go too. A great little project, if only to see how he does the face tracking.

The possibilities of HTML 5 continue to blow me away, and little more than what's offered by noVNC - a VNC (remote desktop protocol) client built using HTML 5 Canvas and WebSockets (with a fallback for browsers that can't use WebSockets). An impressive piece of work from developer Joel Martin.

GCCSense is a standalone program that uses GCC's code analyzers to provide code completion features to any editor of your choice (though Vim and emacs are shown as examples). Beyond knowing that, this demo screencast illustrates how it works beautifully:

Microsoft has unveiled the latest "platform preview" for Internet Explorer 9 and.. long-awaited support for the CANVAS tag has been included! This is a major step forward for IE and one that will, eventually, bring an almost full adoption of HTML 5 standards to the majority of Web users.

The CANVAS element was originally created by Apple to make it possible to build advanced graphic-driven "widgets" in OS X and has become a part of the HTML 5 standard. Safari, Chrome, Opera and Firefox all support CANVAS in their latest versions, but Internet Explorer has held out with its alternative VML system till now. Projects like explorercanvas have emulated CANVAS support in IE but, as of IE 9, should no longer be required.

TechCrunch has a post about today's IE 9 preview and its support of other HTML 5 technologies such as the VIDEO and AUDIO tags.

node.js knockout (probably better called the "Node Knockout", but hey) is a 48 hour coding contest in the same vein as the Rails Rumble - it takes place on August 28-29, 2010. Naturally, the aim is to use node.js, and the contest format is as such:

Build any web project you can imagine;
In 48 hours, starting from scratch;
With a team of up to 4;
Using node.js and whatever additional technology you want

Node Knockout

Thanks to @elliottcable for the lead.

pdf.js is a new JavaScript library by Marak Squires and Matthew Bergman that allows you to create simple PDF files directly in a client's Web browser or, if you choose, with the node.js server-side JavaScript implementation.

Browser-generated PDFs are served using data-based URIs, such as this one (hover over the link to see the long URL). A page has been set up where you can create some on the fly in your browser right now.

Would you believe me if I claimed that an algorithm that has been on the books as "optimal" for 46 years, which has been analyzed in excruciating detail by geniuses like Knuth and taught in all computer science courses in the world, can be optimized to run 10 times faster?

Poul-Henning Kamp

Poul-Henning Kamp (of MD5, Varnish and FreeBSD fame) asks us to think again, if we think we've mastered the art of server performance. In working on the Varnish HTTP accelerator/cache, he's discovered that a supposedly "optimal" algorithm can be improved upon due to the unique side effects of virtual memory systems.

While his article goes into algorithmic complexity, it's remarkably accessible - he has a knack for explaining a dry topic in an enjoyable to read way.