Re: [JGIT] Blame functionality for jgit

>
>> I largely ported the cgit blame algorithm described here
>> https://kerneltrap.org/mailarchive/git/2006/10/12/224187 , the
>> relevant file is builtin-blame.c cf.
>> http://repo.or.cz/w/git.git?a=blob;f=builtin-blame.c;hb=HEAD
>
> OK.  Fortunately Junio has largely given his blessing to this code
> being converted to Java under the BSD license.

> Unfortunately I can't reach incava.org to download java-diff, but
> its entry on sourceforge says it uses an LGPL license.  Within JGit
> we do want to stick to BSD and BSD dependencies, so bringing in
> java-diff as a new dependency isn't something we really want to do.
>
> Fortunately Dscho has been working on a Java diff implementation for
> JGit, and is considering releasing it under a BSD license so we can
> include it.
>
> The interface is still to be decided, but in general we have
> found that running against a byte[] is much faster than running
> against String.  We're probably going to want the diff library to
> take a byte[] and an IntList (as created by RawParseUtils.lineMap)
> and let it work against the byte array ranges directly, without any
> additional allocations, except where it needs to build up its hash
> of records.

> I think eventually we'll have a BSD licensed LCS that will come with
> JGit, which would eliminate the need for such a shim.  I'd like
> to see that happen before blame gets added, but if blame is ready
> and the shim is reasonably well done, I'm inclined to bring blame
> in early.

> - Don't use @author tags, we don't use them anywhere else.
>
> - Please include copyright headers on all new files.
>
> - I think the IDiff, IDifference should be in a ".diff" package so
> we can reuse them for other diff applications.  Especially if we
> wind up using a shim to link to different LCS implementations.
>
> - I think the API for BlameEngine should be more like:
>
>  public class BlameEngine {
>    private final RevWalk rw;
>
>    public BlameEngine(final Repository repo) {
>      this(new RevWalk(repo));
>    }
>
>    public BlameEngine(final RevWalk walk) {
>      rw = walk;
>    }
>
>    public RevWalk getRevWalk() {
>      return rw;
>    }
>
>    public List<BlameEntry> blame(RevCommit c, String path);
>  }
>
>  One reason for this style of API is we can have the engine cache
>  blame records.  This can make a GUI navigator much more efficient
>  as it jumps around between commits and asks for blames over the
>  same data.
>
> - Internally you should *always* use RevCommit, not Commit.
> The RevCommit class can be orders of magnitude faster than Commit.
>
> - Internally you should always use RevTree and TreeWalk to locate
> blob IDs.  Its orders of magnitude faster than Tree and TreeEntry.
>
> - Don't use new String(loader.getBytes()) (e.g. in Origin.getData)
> to get the data for a file.  We want to do raw diffs, so that the
> character encoding is considered as part of the blame.  E.g. if
> a file switches character encodings, the blame has to go to the
> commit that introduced the new encoding.  This is a huge reason
> to use byte[] and IntList over an array of String.
>
> - RawParseUtils.lineMap will be faster than a Pattern of "^(.*)$".
>