sklar.com/blog

One of the books that I'm writing right now has a section on PHP code caches / accelerators. I'd like to include some comparative benchmarks in the book that explore the performance of the different accelerators in various circumstances.

To do the benchmarking properly, though, I need some computers that I don't have, namely a relatively beefy server on a network with a bunch of client computers that can bombard the server with requests. Perhaps someone has a setup like this that I could run some benchmarks on?

I am most interested in the relative performance of the accelerators, not their absolute performance, so the server doesn't have to be anything really beefy, just powerful enough that it doesn't run into problems that are unlikely in the real world (like swapping web server processes to disk when handling five simultaneous requests).

The server should be running Linux (although the presence of a Windows server as well would be a bonus) and the client computers can be running just about anything, as long as they can handle one of the zillion web stress tools out there. The server must be configurable so that while the benchmarks are running, there are no other variable-load programs running (i.e. a mail, DNS, or active web server is no good.) Remote (ssh) access to the machines would be most convenient for me, but if you're in the New York City area, on-site access is OK, too. I can pay you in thanks, acknowledgement in the book, and a free copy or three of the book.

This New York Times article is a neat peek inside the world of Unicode and typography. It also shows how an interest in character encoding can lead to traveling around the world.

Some discussion on a mailing list today about the origin of software CDs and computer books for sale on the streets of Manhattan prompted me to think that if I were a tech-savvy distributor of pirated software for sale (or uploader of pirated software to P2P networks or online distribution sites), I would make these counterfeit CDs of Windows or Office or Dreamweaver MX or whatever also secretly install some spyware that would give me access to the users' machines to sniff data that I could use for financial fraud, or to turn them into spam sending robots, or whatever other malicious purpose I might dream up one day (sell distributed computing services to the criminal underworld?).

There's no need to be really tech-savvy and actually modify the app, just alter the installation script so that it also installs, without notice, the spyware binary somewhere.

Plus, the user wouldn't have much recourse. Some would disable or delete the spyware, sure, but the users who can't/dont aren't going to get their money back from wherever they bought it (if they bought it at all) and they can't call up Microsoft or Macromedia or whomever and complain about it.

I suppose a conspiracy minded individual could think that such infected pirated software count be introduced into the wild by software companies to discourage piracy. A more malicious version of polluting filesharing networks with files that look like popular songs, but aren't (which I think is great, BTW).

I've had this blog for a few weeks and in the same time period, I've been experimenting with some RSS aggregators. This means that every hour, my computer makes a request to ten or twelve servers to get RSS feeds and lots and lots of computers make requests constantly to my web server for my RSS feed.

This strikes me as wasteful and inefficient. What if, instead, a blog's RSS feed was treated like a newsgroup and hierarchically distributed? An ISP or a company or the same kinds of folks that run NNTP servers could run RSS servers which would periodically grab RSS files from individual blogs (or other upstream servers) and then pass on the changes to downstream servers and readers.

Blog URLs are easily turned into general-to-specific names for organization in this newsgroup-like world. Just invert the components of the hostname in the URL and keep the pathname as is: www.sklar.com/blog becomes com.sklar.www/blog. (Which I wish is how URLs were organized to begin with, but that's another story.)

Even better, instead of individual blog/NNTP servers getting the RSS file from a blog, the blog can upload it to a server or a set of servers to get the process rolling. I know there are some ping services now that monitor changes, so that's a step in the right direction.

Adam, Chris, and I will be talking about PHP at the NYSIA Open Source SIG meeting on October 7.

I am teaching "Introduction to PHP Programming" again this fall at the New School University. The class will meet once a week on Wednesday evenings, beginning on September 17. If you're in (or near) New York City, this is a great way to dive in to PHP. The first class starts with the basics of language syntax and by the end of the semester, you'll be able to build a database-based web site.

You can get more info and sign up here:
--> http://kaplan.dialnsa.edu/register/index.cfm?deptcode=NCIC

My preferred SSH client for Windows is PuTTY. One of the reasons why I like it is that it comes with an authentication agent, Pageant. When Pageant starts up, it loads up my SSH keys and prompts me for their passphrases. When I use PuTTY to SSH somewhere, if Pageant has an authorized key, it jumps in and supplies the right info so I don't have to enter a password or a key phrase.

For storing passwords and other goodies, I use Password Safe. It's a "vault" that stores information in a Blowfish-encrypted, passphrase protected file. When I need a password, I can run Password Safe and grab it.

While each of these tools are handy on their own, I would be even happier if I could put my SSH keys into Password Safe and have it do the authentication agent magic that makes Pageant so swell. I have very little experience with Win32 or MFC programming, so perhaps this would be a good project to learn on. Unless, among the hordes of sklar.com/blog readers, there is a Win32 whiz that wants to try...