Developer forums (C::B DEVELOPMENT STRICTLY!) > CodeCompletion redesign

Question about the code completion revamping done around 2010...

<< < (5/5)

ollydbg:

--- Quote from: rickg22 on November 07, 2011, 06:44:37 am ---
--- Quote from: ollydbg on November 07, 2011, 06:08:03 am ---
We don't have a word "ded"? or "dlength,"?


--- End quote ---

needed, wordlength. What you're seeing aren't words, but tree vertices.

--- End quote ---
Sorry, I still no follow you, e.g. see the tree below:

--- Code: ---- "" (0)

      [skip..]

      +- "d" (158)
      |         +- "ed" (267)
      |         +- "i" (299)
      |         |         +- "fication" (159)
      |         |         \- "ng" (300)
      |         +- "length," (253)
      |         \- "o" (11)

--- End code ---
A full key is the string that accumulate all the labels from the root to the leaf, here the "d"(158) is a child of the root ""(0), so I can see the full string is "" + "d" + "ed" => "ded".

Am I right?

MortenMacFly:

--- Quote from: ollydbg on November 07, 2011, 08:20:05 am ---Am I right?

--- End quote ---
If it would be the tree's root, yes. But I guess Rick showed the tree only partially and there are simply missing some parent nodes.

rickg22:
Oh crap, I think i screwed the tree algorithm. I need to recopy the old version then :P

EDIT: Oh, no. I saw what happened. The tree I displayed is a SUFFIX tree. It contains all the suffixes of the dictionary for faster lookup. Anyway, I hadn't implemented it well and it got on an endless loop after processing a big file. I'm rewriting the algorithm to use a less complicated edge labeling scheme, and possibly make some other modifications.

Navigation

[0] Message Index

[*] Previous page