Re: bug in org-store-link

[Scott Otterson <scotto@u.washington.edu>]

Yeah, I guess that instead of saying it was a small bug, I should have 
said that it's a bug of small consequence (for most users, but matters 
to me, least). 

The ambiguity problem you mention could be solved by matching more than 
one line.  To keep the string stored in the org link short, 
org-store-link could expand it to another line only when needed for a 
unique match.  Or, it could expand just enough _words_ to ensure 
uniqueness, plus maybe one word on each end for some insurance against 
future changes.

Future changes are the harder part.  In speech recognition, there's an 
analogous problem where there's a need to match a transcript to 
recognized speech, which may have a lot of word errors, insertions and 
deletions.  The simplest solution commonly employed is a word-level 
Levenshtein distance:

    http://en.wikipedia.org/wiki/Levenshtein_distance
    (this is for chars, but you get the idea)

Scott

Carsten Dominik wrote:
> Hi Scott, this is not a small bug, but a problem that is really hard 
> to solve.
> Supposed I used the exact line text to search, then you still have two 
> lines in the buffer
> that would match.
>
> This is really about what strategy should be used to find a location
> in a file that has possibly changed.
> I have no good answer to that.  Do you?
>
> - Carsten