Sometimes software engineers forget the importance of history. I often sit in meetings at Yahoo! when someone goes on a rant about how bad some of the legacy Yahoo! technology was. I’m referring to HF2K and another technology known as idaho.
Which were written almost completely in house and are the equivalent of PHP today.
What I often remind them was that back in 1996, the game was totally different. Funding was often very tight, competition was fierce, the web was so new, the scale and pace of growth was so enormous and there was a lack of proven, extentisible programming languages.
Also the landscape changed so much, a web company need complete total control over it’s core technology and not be dependent on any third party vendor.
All of these factors plus the simple fact that there was a huge pool of kickass C programmers who could write very low level web server code. I challenge you to find more than one programmer in your group who ever wrote an apache plugin in C.
I wrote these C plugins for news, even though they were hard, when they were lauched, you can bet that I’ve tested them thoroughly and I’ve understood the entire web server interaction, most of the HTTP protocol that could affect the plugin.
Even today idaho out performs anything out there. But because machines are cheaper and faster, performance is no longer a factor.
Remember fellow engineers, when you sit back and criticize, put yourself back in 1996, imagine if Yahoo! picked Java servlets, or even PHP or mod_perl, would there be a Yahoo! today? I bet not!
I encourage criticism in my engineering group, but when you criticize, it’s yours to fix. hee