Blogging Like a Hacker | Hacker News Reader

Blogging Like a Hacker

tom.preston-werner.com

55 pointsmojombo33 comments

Programmer writes own website platform. Standard.

I've often thought that history etc for cms/blog systems should be pushed off to a proper version control system etc. I even drafted out a project spec to implement basically the same thing as this guy. Then I put it to the very bottom of my priority list. Emacs muse-mode + darcs FTW.

edit: not to sound like a prick - good on the author for working on a personal project, getting something he likes working and then using it. I just don't think it's news.

"... good on the author for working on a personal project, getting something he likes working and then using it. I just don't think it's news ..."

Neat software hacks are always news. I couldn't think of a better place to post your new bits of code.

He implements this and open sources it, and yet it is not worth a post on HN to you? What is?

"This guy" would be interested in seeing your blog. I've not seen what Emacs Muse can do and would love to see an example.

I was sorely tempted to follow the same path. But then I had an insight: github may be a terrible blog platform, but it is _good enough_ by itself.

So far, it sucks not having full text RSS feeds and having to manually tag posts in delicious. It's slow and some readers seem to miss the comments. But a Tom says, you get to write in textile or markdown. And as for not having CSS, I have convinced mself it's a benefit; I am not spending any time tweaking page layout.

http://tinyurl.com/homoiconic

On the topic of blogs and CMS...

Does anyone else think that Wordpress is getting too complex with all the added features due to it being open source?

I feel like people like my dad, who would be happy with a very straightforward and simple-to-use interface without all the added features, are turned off by the complexity of Wordpress.

I think a very lightweight CMS that would basically be a password protected AJAX textarea that could easily be embeded into any template would be ideal. No back-end interface or complex editing tools.

It seems almost inevitable that software projects become more complicated over time. Unless simplicity is a core requirement, complex features are constantly added to popular software (and rarely removed). By trying to solve a wide variety of problems, the number of options and screens must correspondingly increase. I'm really impressed by new approaches like Posterous that challenge the way blogging has been done in the past. It's these kinds of approaches that can lead to software that can be either as simple or as complex as the user needs them to be.

I agree, Posterous is awesome. It would be even more awesome if you could somehow just embed a little code on your own template in order to display your blog on your own design.

As far as I know, you currently have to learn the platform (Wordpress, MovableType, etc) in order to make a custom design. I feel like there needs to be a way to have the blog fit into the design without having to code your own CMS.

They haven't been around very long, I wouldn't be surprised if they figure out a slick way to address those problems in the future. I'll be interested to see how they approach it if they do.

Thanks for the support, guys. This is a great idea and definitely something we want to get to.

You may be looking for Drupal. The core is extremely light and you could easily configure it to do just that. Drupal's done a great job of only adding the most necessary features. I'd like to see them pull Book + Forum and add CCK + Views to core, but I think that they still need to keep some obvious out-of-the-box functionality for new users.

I remember when I chose Drupal. I downloaded the code of five open source CMSs and finally decided to use Drupal, mostly because the code was the shortest and most legible. It has grown quite a bit since then, but I would still recommend it.

Thanks, I'll look into it.

I'm not totally clear on one of your arguments. Why would it become more complex simply because it's open source?

You may also be interested in:

* chronicle blog compiler: http://www.steve.org.uk/Software/chronicle/

* ikiwiki wiki / bliki compiler: http://ikiwiki.info/

I find ikiwiki particularly interesting in that it coordinates with a local vcs (git, hg, svn, etc) to handle revisions and history.

Another good static site generator is http://webby.rubyforge.org

It features support for LaTex, markdown, textile filters, autobuild, publishing only changed pages etc. Very small and clean.

'gem install webby' is all needed to get you started.

Webby rocks.

I'm all for people scratching an itch and rolling their own, but I'm curious why Webby wasn't suitable for the task.

In fact, this:

http://github.com/mojombo/tpw/tree/master/index.html

looks exactly like what Webby does.

Webby is a great system and very close to what I needed; indeed I took many ideas from it. But webby does not help me with the blog specific parts that I wanted to have automated. The entire point of Jekyll is to remove every last bit of pain associated with creating, maintaining, and posting to my weblog.

Webby has a blog mode, though I've not used it, but I'm guessing it wasn't quite what you needed.

When I see people sort-of reinvent a wheel or two I wonder what the reason was.

Out of curiosity, did you consider forking Webby, or offering to add in the parts you wanted back to the Webby project? Is it easier to grab the ideas you like and start fresh on your own than to try to hack on someone else's existing code?

For myself, when I find myself writing something similar to an existing project, it tends to be because a) I wasn't quite aware of what the other project offered, b) (more often) the 80/80 rule kicks in: The existing code gives me 80% of what I want, but I'll be spending 80% of my development time trying to get the rest of what I need, and writing my own version looks to be more efficient in the long run. This is more so if I expect to be reusing that code for many projects, so I can readily see the motivation in creating your own blogging software if you plan on using the hell out of it, regardless of what existing tools do.

Jekyll is a re-implementation of existing projects except they don't have version control. I use rassmalog, which is small and easy to hack on, and supports plugins. There is also rog, and hobbix, both of which I looked at using but have not blogged with.

Tom, that was very inspiring.

Coincidentally, I decided to roll my own blog a week ago.

It's written in Lisp.

http://jamrides.com/blog/s/20081114004158.html

I enjoyed reading your blog, I especially liked the "And nothing big starts big". quote in your about page. Intersting startup too!

Very nice.

It looks surprisingly close to the blog system I had stewing in my brain. But hey... now I can just hack at your's. Only thing you're missing is comments, but perhaps that's what HN is for...

Now I just have to learn Ruby.

I've opted to rely on HN for discussion in lieu of comments, but you could easily hook up disqus or another embedded commenting system. Send me a pull request on GitHub when you've got your features hacked in!

I almost created the exact same thing when I started my blog. It was a source-controlled (SVN, though) Textile powered static blogging engine.

Comments were the reason I ended up looking into wordpress and movable type. I thought heavily about autosubmitting every post to reddit or hacker news and using that as the discussion but that could be constituted as spam (or offtopic). I thought about using one of the out-sourcing services like disqus or JS-KIT. The problem (back then) was they didn't index properly on search engines and it has really really high switching costs.

If those two issues have been solved (the search engine one might have been), I'd consider switching to a static engine.

For most articles I choose to read from my RSS, I pull up the corresponding HN page. The discussion here is a lot more fruitful than any I can find on a blog.

As sites grow, re-generating and re-copying the entire site may get cumbersome and time consuming. An obvious feature is the make-like ability to determine what has changed since the last invocation, and only modify those files. You can accomplish pushing newly generated sites to remote servers without undue copying through rsync, or an approximation of it.

Yeah, there's lots of smarts left to add to Jekyll. Only regenerating modified posts/assets is a big one.

It might be easier to actually use make instead of reproducing its functionality. This implies that an invocation of jekyll would produce one static page instead of the whole site.

On the subject, to the people who have rolled their own similiar blogs: What text formats are you using?

Many seem to follow the "title: \n date: \n tags: \n" type style which seems to do the job pretty well (and can be easily parsed).

Any other formats worth investigating?

I enjoy his writing, and enjoy reading about ideas for generating sites, but I'll use my own blog compiler, thanks.

After I implemented the world's best photo-sharing site using Scheme I discovered it was also a decent blogging platform. See my blogs here: http://friendfeed.com/brlewis?service=blog

For the lulz: http://github.com/cliffmoon/fake-tpw/commit/2f5a0c888c784a9a...

http://github.com/cliffmoon/fake-tpw/commit/df2fbc653e928624...

http://github.com/abhay/fake-tpw/commit/f81f8577111f6e82cdee...

Reply on news.ycombinator.com