Re: Spell checking identifiers

> [There's a vast amount of work on edit distance.  My guess is they
> use something like Levenshtein, but rather than use a constant
> distance of 1 between different letters, the distance varies depending
> on how different the letters look. -John]


This clang blog specifically mentions Levenshtein,




http://blog.llvm.org/2010/04/amazing-feats-of-clang-error-recovery.html#spell_checker


and it looks like what people do is to go through the entire symbol
table and compute it against the individual erroneous identifier.


I thought that'd be a bit on the expensive side, because C++ files
can have 100k+ (or millions?) of lines after preprocessing, so one
translation unit really can go up to million identifiers in practice.
[I don't know if that actually happens but I don't think it's safe
to assume it doesn't.]


In the 10 years since, people may have changed from standard Levenshtein
as you mention.


But then, maybe compilation speed for erroneous input isn't really
important. rustc is slow for a short input file in both cases [which
could be the startup cost.]


--
Johann | email: invalid -> com | www.myrkraverk.com/blog/
I'm not from the Internet, I just work there. | twitter: @myrkraverk

Return to the comp.compilers page.
Search the comp.compilers archives again.