[jgit-dev] WorkingTreeIterator - the writing side

[Christian Halstrick <christian.halstrick@xxxxxxxxx>]

Hi,

by having different implementations of WorkingTreeIterator we allow
jgit-users to define how content in the working tree is accessed. This
works great for the reading side - but the writing side is not covered
yet. How do we allow jgit-users to specify how working tree content
should be written. When jgit is used on server-side (gerrit) java file
I/O may not be optimal.

What do you think: is WorkingTreeIterator also a good place to have
writing abstractions (e.g. by adding a openEntryOutputStream()
method)? Don't we get into trouble when in the middle of a walk
somebody starts writing new content? We have to support deletion and
creation of files also.

Ciao
  Chris