Finding Things

[[deleted user]]

A little whining for this chapter... It started off just fine, with a sales pitch for Ruby (and dynamic languages in general) and regular expressions and hash tables. I was excited to see some performance analysis and discussion about scale issues, and then I lost the thread. The author seemed to first focus on the excessive load time for his web log analysis program, especially compared with the negligible search time. Then he offers up the ultimate Java binary search solution, which does nothing to help the load time and instead focuses on the part of his solution that doesn't appear to need the help (but should help with the memory footprint of the overall solution). And, after providing detailed performance data, he provides none for his final solution. The author finishes with a very terse and almost flippant description of the issues involved in doing a Google-scale search.

It is pretty interesting to see Java offered up as a performance solution!

This chapter needed another round of editing.

[elevel]

I agree. Poor topic cohesion in this chapter.

Plenty to cover involving just regular expressions alone - perhaps across multiple languages.

[Erik]

I found some of the previous chapters a little "preachy" too, so that didn't surprise me much in this chapter.

The "everything is a text file" view simplifies this issue a lot too. At the beginning of the chapter, I was hoping the direction was headed toward "finding things" in other types of files (Office Documents, Multimedia, PDF, etc...). I thought it might go in a SQL direction too, but it didn't.

The table discussion was pretty basic, but definitely important information. I was interested to see how this scales too. Although, it does mention partitioning and distribution. A search time of a few GB and minutes is pretty small scale.