(Note: I'm ccing ruby-dev@lucene.apache.org since the Ruby people are
probably going to have to think about these same issues, so getting
their opinion seems to make sense.)
The current version of Lucene4c in Subversion is based off of the
release version of Lucene 1.4.3. There was some previous work put into
working with the current development version of Subversion, but it ran
into some trouble. It's likely that the problem was related to building
the Java code with a 1.5.x JDK, which doesn't play nicely with GCJ.
At some point, we're going to have to transition to a newer version of
Lucene, the question is simply when.
Things to keep in mind:
1) Newer versions of Lucene add new features, which might be useful.
2) Newer versions of Lucene deprecate interfaces, some of which we are
actually using (the Field interface is the big one here), so getting the
change over with soon would mean we wouldn't end up with people
depending on interfaces that will eventually go away.
3) It's unclear when newer versions of Lucene will be actually released,
so if we switch now it's unclear when we'll have a non-moving target to
work with.
Personally, I think it isn't that big a problem to stick with 1.4.x for
now, since we have an extremely limited number of users anyway. But
before we actually release anything we should probably at least make an
effort to make the transition, to avoid having people depend on
interfaces that are deprecated.
Does anyone have any other opinions on this topic?
-garrett
probably going to have to think about these same issues, so getting
their opinion seems to make sense.)
The current version of Lucene4c in Subversion is based off of the
release version of Lucene 1.4.3. There was some previous work put into
working with the current development version of Subversion, but it ran
into some trouble. It's likely that the problem was related to building
the Java code with a 1.5.x JDK, which doesn't play nicely with GCJ.
At some point, we're going to have to transition to a newer version of
Lucene, the question is simply when.
Things to keep in mind:
1) Newer versions of Lucene add new features, which might be useful.
2) Newer versions of Lucene deprecate interfaces, some of which we are
actually using (the Field interface is the big one here), so getting the
change over with soon would mean we wouldn't end up with people
depending on interfaces that will eventually go away.
3) It's unclear when newer versions of Lucene will be actually released,
so if we switch now it's unclear when we'll have a non-moving target to
work with.
Personally, I think it isn't that big a problem to stick with 1.4.x for
now, since we have an extremely limited number of users anyway. But
before we actually release anything we should probably at least make an
effort to make the transition, to avoid having people depend on
interfaces that are deprecated.
Does anyone have any other opinions on this topic?
-garrett