Bug #6525: misleading example in Permformance section of Regexp documentation?

• Status: Closed
• Assignee: drbrain (Eric Hodel)
• Target version: -
• ruby -v: 1.9.3

[nathanst (Nathan Stratton Treadway)]

In the Performance section of the Regexp documention, it is stated that changing from Regexp.new('a?' * 29 + 'a' * 29) to Regexp.new('\A' 'a?' * 29 + 'a' * 29) speeds up the match significantly. However, '\A' 'a?' * 29 actually expands to "\Aa?\Aa?....\Aa?" -- which doesn't seem like it's what you'd actually want, since [for example] Regexp.new('\A' 'a?' * 29 + 'a' * 29).match('a' * 50) will only match 30 'a' characters [the 29 from 'a' * 29, plus one single one from the set of '\Aa?' items. [That is, Regexp.new('\A' 'a?' * 29 + ...) is doesn't match any more 'a' characters than Regexp.new('\A' 'a?' + ...) would.]

One can get around this by changing the expression to Regexp.new('\A' + 'a?' * 29 + 'a' * 29), in which case additional "a" characters do get matched as the string gets longer... but with that Regexp the .match('a' * 29) still takes about the same amount of time as the original one.

So I wonder if there is some better example that could be used here?

Nathan

[drbrain (Eric Hodel)]

• Category set to doc
• Assignee set to drbrain (Eric Hodel)

[drbrain (Eric Hodel)]

• Status changed from Open to Closed
• % Done changed from 0 to 100

This issue was solved with changeset r35862.
Nathan, thank you for reporting this issue.
Your contribution to Ruby is greatly appreciated.
May Ruby be with you.

---

• doc/re.rdoc (Performance): Replaced incorrect example of reducing
  backtracking through anchoring with reduced backtracking through a
  range. [ruby-trunk - Bug #6525]