Aaron N. Tubbs bio photo

Aaron N. Tubbs

Dragon chaser.

Twitter Facebook Google+ LinkedIn Github

I’ve complained, in many cases, offline in the past about PHP. I want to preface all of this with a few things:

  • I’ve used PHP, professionally, and made good money from doing so. I enjoyed the work when I did it.
  • PHP, as a web scripting, rapid development, and prototyping engine, is just about perfect for the purpose.
  • PHP, as a development environment for internal tools, administrative aides, and so forth, is just about perfect for the purpose.
  • My meager website runs in PHP. I can’t hate it that much!

Right, where was I. Oh, yeah … PHP attracts a bunch of idiots because it’s easy. It’s a lot like ASP. There is not an intrinsic flaw in the language itself that makes it so terrible (well, with ASP that’s a little more true, but now I’m just being a syntax snob) … it’s just one of those systems where they’ve made a lot of functionality accessible for the average moron, which leads to people developing all manner of extremely poor-architected solutions, in terms of complexity, scaling, and so forth. On the other hand, it’s brought web development and programming to a lot of people for which it would be inaccessible otherwise, which is a really good thing, as it’s gotten a lot more ideas, design, and content out there that would not exist otherwise. So really, I’m glad PHP happened, I’m just not real happy with all (most?) of its users, as many tend to be the sort that have no concept of scale, optimization, safety, security, or complexity.

In any event, where I’m going with all of this is there’s this newfangled log analyzer called mint, which everybody is calling the cat’s meow. As far as I can tell, it doesn’t really offer anything over any other log analysis package, based on its current feature description and demo. What it does provide is:

  1. It looks pretty.
  2. It doesn’t require the analysis of regular logfiles, because it makes its own. The advantage to this is it can gather things not normally gathered or figured out in Apache logfiles and so forth.
  3. It looks pretty.
  4. It’s easy to use.
  5. Oh my god, it’s beautiful! I want one! The tables are all dressed up and look better than my blog, and OMG look, my referers are all pretty too! I should stop using words like “babies” and “eating” in the same blog entries, because I’m attracting some really strange search results on google and OMGWTFBBQ MY BLOG IS COOL AND POPULAR AND PEOPLE LOVE ME!

Really, that’s it. I can’t find anything that would suggest any other functionality than that, but people are drawn to flashing lights and pretty. So, where does this all become a problem? It becomes apparent that this thing may suffer somewhat in architecture. FIrst off, it adds additional logging to the server (well, of course we’re a LAMP project, and that means more mysql, another stellar technology that due to its ease of use is most often abused) every time a page is hit, rather than working with existing logfiles (or even extending existing logging mechanisms; since this thing is pretty tightly tied to a platform already, one may as well hack up Apache’s logging mechanisms instead, to get what they need, but that’s hard). Next, it requires the client to read and support javascript. This is great, because it does a nice job of stopping referer spam, but it also means one has to trust in their visitors to have javascript and use it. Third, it’s pretty tightly tied to a specific site, and has no real intrinsic ability to scale across multiple domains in an integrated fashion … and the FAQ is full of “uh, is this thing actually useful?” questions related to this.

All in all, it’s an interesting academic exercise, but it’s completely pointless for anything but the most unpopular of weblogs and personal sites. Granted, I fall into that category, so maybe mint is for me, but reading things like this just kill me:

The demo has been temporarily disabled. While Mint can easily handle recording hundreds of hits a minute, it is not optimized to display data at that rate. Please check back once some of the initial interest has died down.

What a pile of steaming horse shit! Nevermind that it’s poor at displaying its information at speed (I’m heading into speculation now, but I’m guessing lots of post-processing work in PHP, rather than just gluing some data/tools together), but it says the problem right there … “can easily handle recording hundreds of hits a minute.” Hundreds of hits a minute? Good heavens, I hope so! Hundreds of hits a minute is not a big number on the web!

So getting back to what I’m talking about, I’m not suggesting Shaun Inman is an idiot. Quite the contrary. He’s a genius at marketing and business, and I think he’s going to make a killing on this thing, because it’s easy, pretty, cheap, and succinct. He has identified a niche — small blog authors searching for ego padding — and I think he’s doing a swell job of producing a product perfectly targeted for that niche. But, his tool seems to exhibit the classic flaw of the rapid development LAMP crowd — in the form of a complete inability to scale and perform under pressure, or leverage existing technologies more effectively.

I’ll add the disclaimer that this is all based on external evidence, as I have not seen the internals of this thing, so everything I’m saying could be outright wrong, but I guess I’m just having a hard time seeing the value in this, and why everybody is jumping for joy. Has the corporate world blinded me by forcing me only to think in terms of long-term usability and scale? Would google make a tool like this? Am I missing the point?