I've decided to finally document my web server permission setup tips in a central place, and what better place than my blog! This technique diverges slightly from best practice in that I don't create a separate group (such as
www-pub). I simply use the existing
apache group. So for reference, here are my notes.
I've had a string of unfortunate data losses recently. As far as I could tell, it was just bad luck (nothing malicious), but frustrating nonetheless. The really sad part is that I thought I had a pretty decent backup solution in place, yet still lost data.
After having experienced this type of data loss on a few occasions in a short period of time, I've had a chance to test the resiliency of my backup solution.
I've had some success running WordPress (3.3) on an EC2 micro instance. After much research into tuning a LAMP stack under a low memory environment (EC2 micro only carries 613MB of memory), I believe I have a pretty solid setup going.
Note that these instructions assume that you are creating a server dedicated to a single installation of WordPress (i.e. no virtual hosts).
I've started playing around with Meteor and it's been delivering as advertised. I'm very excited about what it means for the web industry. It feels like a massive, huge step forward.
I was self-brain-dumping some thoughts concerning RESTful APIs and those thoughts progressively turned into rants. I decided to shelve that blog post for the time being, but with the advent of Meteor (and Derby alike), I think my rants aren't necessarily unfounded after all.
I thought I invented the technique known as Responsive Design sometime mid last year (2011). Then one of my co-workers came across an article on alistapart showing the same technique, pre-dating me over a year... oh well, so it seems I was on the right track at least! And the term "Responsive Design" sure sounds better than "MOB Layout".
It struck me recently that the conventional approach to loading libraries like jQuery through CDNs (content-delivery networks) might be a bad idea.
I've always implemented deep-links without question, because I come from a web site (vs web app) background where indexing is very important.
When I dove into writing web applications, it seemed just as prevalent a concept. It felt familiar so I just went with it.
But after actually using my web application on a regular basis (and seeing others interact with it), I wonder if I did the right thing in implementing deep-linking...
With HTML5 quickly becoming the de facto markup language for the web, we front-end developers are now armed with a slew of semantic tags to help better describe the meaning of various parts of text on our web pages. This is a big step in the right direction.
But we're still screwed... because of CSS.
From all the research and testing I’ve done of the xPath
lang() function, I can only conclude that it is broken. Its behavior does not make sense to me. Perhaps my expectation of its functionality is wrong because I’m using it in an unintended way.