This is a start to fixing the ObjectLoader save during repack
problems I talked about yesterday. It applies on top of the
WindowedFile+PackFile patch I sent earlier today.
Patch 1 and 2 are trivial.
Patch 3 fixes the broken 10/10 test case I raised yesterday.
Patch 4 introduces that test case.
Patch 5 starts to fix PackWriter to be safe during repack.
Its still a bit iffy. The basic logic here is sound, but I'm more
worried about someone else calling PackFile.close() and closing the
underlying RandomFile while we are otherwise accessing the file
during the copyRaw in PackWriter. I *think* its impossible for
that to occur, but I haven't proven it to myself yet, and its late,
so this one is RFC for now.
Shawn O. Pearce (5):
Remove throws IOException from UnpackedObjectLoader.getCachedBytes
Add missing @Override annotations to UnpackedObjectLoader
Fully materialize an ObjectLoader before returning it from
ObjectDatabase
Test that ObjectLoader stays valid across repacks
Teach PackWriter to recover from removed/replaced packs
.../org/spearce/jgit/lib/ConcurrentRepackTest.java | 40 +++++++
.../jgit/lib/DeltaOfsPackedObjectLoader.java | 7 +-
.../spearce/jgit/lib/DeltaPackedObjectLoader.java | 41 +++----
.../jgit/lib/DeltaRefPackedObjectLoader.java | 9 +-
.../src/org/spearce/jgit/lib/ObjectDirectory.java | 3 +-
.../src/org/spearce/jgit/lib/ObjectLoader.java | 23 +---
.../src/org/spearce/jgit/lib/PackFile.java | 14 +-
.../src/org/spearce/jgit/lib/PackWriter.java | 118 +++++++++++++-------
.../org/spearce/jgit/lib/PackedObjectLoader.java | 78 +++++++++++--
.../org/spearce/jgit/lib/UnpackedObjectLoader.java | 4 +-
.../spearce/jgit/lib/WholePackedObjectLoader.java | 21 ++--
.../src/org/spearce/jgit/lib/WindowCache.java | 59 +++++++----
12 files changed, 280 insertions(+), 137 deletions(-)
--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html