The new Add command adds files to the Git Index.
It uses the DirCache to access the git index. It
works also in case of an existing conflict.
Fileglobs (e.g. *.c) are not yet supported.
The new Add command does add ignored files because
there is no gitignore support in jgit yet.
Bug: 318440
Change-Id: If16fdd4443e46b27361c2a18ed8f51668af5d9ff
Signed-off-by: Stefan Lay <stefan.lay@sap.com>
After refactoring ReadTreeTest the tests failed for filesystems
with coarse modification time granularity. This is fixed by
explicitly telling the repo to reread the index after we build
a new index.
Additionally the test testDirectoryFileSimple was simplified
by using buildTree() instead of misusing GitIndex to construct
trees.
Change-Id: I20d2f097491e4cc8c657a696beabc7026b485017
Signed-off-by: Christian Halstrick <christian.halstrick@sap.com>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
This patch adds ignore compatibility to jgit. It encompasses
exclude files as well as .gitignore. Uses TreeWalk and
FileTreeIterator to find nodes and parses .gitignore
files when required. The patch includes a simple cache that
can be used to save results and avoid excessive gitignore
parsing.
CQ: 4302
Bug: 303925
Change-Id: Iebd7e5bb534accca4bf00d25bbc1f561d7cad11b
Signed-off-by: Chris Aniszczyk <caniszczyk@gmail.com>
Signed-off-by: Stefan Lay <stefan.lay@sap.com>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
I added test cases to cover the majority of the code. It's not 100%
coverage yet, but the remaining bits are small.
Change-Id: Ib534c8e94b13358b8b22cf54e2ff84132bae6d14
Optimized a small loop in findExactRenames. The loop would go through
all the items in a list of DiffEntries even after it already found
what it was looking for. I made it break out of the loop as soon as
a good match was found.
Change-Id: I28741e0c49ce52d8008930a87cd1db7037700a61
The javadoc for the setRenameLimit method in RenameDetector said
that you could only have limits in the range (0,100), implying
that 0 and 100 were illegal inputs. The code, however, allowed 0 and
100. I changed the javadoc to say that the range [0,100] was legal.
I also documented the IllegalArgumentException that is thrown if the
limit is outside that range.
Change-Id: I916838f254859f6f0e1516bb55b8e7dc87e57dc2
The scoring method was not taking into account the similarity of
the file paths and file names. I changed the metric so that it is 99%
based on content (which used to be 100% of the old metric), and 1%
based on path similarity. Of that 1%, half (.5% of the total final
score) is based on the actual file names (e.g. "foo.java"), and half
on the directory (e.g. "src/com/foo/bar/").
Change-Id: I94f0c23bf6413c491b10d5625f6ad7d2ecfb4def
The scoring logic in SimilarityIndex was dividing by the max file
size. If both files are empty, this would cause a div by zero
error. This case cannot currently happen, since two empty files
would have the same SHA1, and would therefore be caught in the
earlier SHA1 based detection pass. Still, if this logic eventually
gets separated from that pass, a div by zero error would occur.
I changed the logic to instead consider two empty files to have a
similarity score of 100.
Change-Id: Ic08e18a066b8fef25bb5e7c62418106a8cee762a
Prior to this change, files that were very different in size (enough
so that they could not have enough in common to be detected as
renames) were still having their scores calculated. I added an
optimization to skip such files. For example, if the rename detection
threshold is 60%, the larger file is 200kb, and the smaller file is
50kb, the pair cannot be counted as a rename since they cannot
possibly share 60% of their content in common. (200*.6=120, 120>50)
Change-Id: Icd8315412d5de6292839778e7cea7fe6f061b0fc
The pack.windowlimit configuration parameter places an upper bound
on the number of bytes used by the DeltaWindow class as it scans
through the object list. If memory usage would exceed the limit
the window is temporarily decreased in size to keep memory used
within that bound.
Change-Id: I09521b8f335475d8aee6125826da8ba2e545060d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
If we have multiple CPUs available, packing usually goes faster
when each CPU is assigned a slice of the available search space.
The number of threads to use is guessed from the runtime if it
wasn't set by the caller, or wasn't set in the configuration.
Change-Id: If554fd8973db77632a52a0f45377dd6ec13fc220
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
PackWriter now caches small deltas, or deltas that are very tiny
compared to their source inputs, so that the writing phase goes
faster by reusing those cached deltas.
The cached data is stored compressed, which usually translates to
a bigger footprint due to deltas being very hard to compress, but
saves time during writing by avoiding the deflate step. They are
held under SoftReferences so that the JVM GC can clear out deltas
if memory gets very tight. We would rather continue working and
spend a bit more CPU time during writing than crash due to OOME.
To avoid OutOfMemoryErrors during the caching phase we also trap
OOME and just abort out of the caching.
Because deflateBound() always produces something larger than what
we need to actually store the deflated data, we copy it over into
a new buffer if the actual length doesn't match the buffer length.
When packing jgit.git this saves over 111 KiB in the cache, and is
thus a worthwhile hit on CPU time.
To further save memory we store the inflated size of the delta
(which we need for the object header) in the same field as the
pathHash, as the pathHash is no longer necessary by this phase
of the packing algorithm.
Change-Id: I0da0c600d845e8ec962289751f24e65b5afa56d7
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
PackWriter now produces new deltas if there is not a suitable delta
available for reuse from an existing pack file. This permits JGit to
send less data on the wire by sending a delta relative to an object
the other side already has, instead of sending the whole object.
The delta searching algorithm is similar in style to what C Git
uses, but apparently has some differences (see below for more on).
Briefly, objects that should be considered for delta compression are
pushed onto a list. This list is then sorted by a rough similarity
score, which is derived from the path name the object was discovered
at in the repository during object counting. The list is then
walked in order.
At each position in the list, up to $WINDOW objects prior to it
are attempted as delta bases. Each object in the window is tried,
and the shortest delta instruction sequence selects the base object.
Some rough rules are used to prevent pathological behavior during
this matching phase, like skipping pairings of objects that are
not similar enough in size.
PackWriter intentionally excludes commits and annotated tags from
this new delta search phase. In the JGit repository only 28 out
of 2600+ commits can be delta compressed by C Git. As the commit
count tends to be a fair percentage of the total number of objects
in the repository, and they generally do not delta compress well,
skipping over them can improve performance with little increase in
the output pack size.
Because this implementation was rebuilt from scratch based on my own
memory of how the packing algorithm has evolved over the years in
C Git, PackWriter, DeltaWindow, and DeltaEncoder don't use exactly
the same rules everywhere, and that leads JGit to produce different
(but logically equivalent) pack files.
Repository | Pack Size (bytes) | Packing Time
| JGit - CGit = Difference | JGit / CGit
-----------+----------------------------------+-----------------
git | 25094348 - 24322890 = +771458 | 59.434s / 59.133s
jgit | 5669515 - 5709046 = - 39531 | 6.654s / 6.806s
linux-2.6 | 389M - 386M = +3M | 20m02s / 18m01s
For the above tests pack.threads was set to 1, window size=10,
delta depth=50, and delta and object reuse was disabled for both
implementations. Both implementations were reading from an already
fully packed repository on local disk. The running time reported
is after 1 warm-up run of the tested implementation.
PackWriter is writing 771 KiB more data on git.git, 3M more on
linux-2.6, but is actually 39.5 KiB smaller on jgit.git. Being
larger by less than 0.7% on linux-2.6 isn't bad, nor is taking an
extra 2 minutes to pack. On the running time side, JGit is at a
major disadvantage because linux-2.6 doesn't fit into the default
WindowCache of 20M, while C Git is able to mmap the entire pack and
have it available instantly in physical memory (assuming hot cache).
CGit also has a feature where it caches deltas that were created
during the compression phase, and uses those cached deltas during
the writing phase. PackWriter does not implement this (yet),
and therefore must create every delta twice. This could easily
account for the increased running time we are seeing.
Change-Id: I6292edc66c2e95fbe45b519b65fdb3918068889c
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
This is a horribly crude application, it doesn't even verify that
the object its dumping is delta encoded. Its method of getting the
delta is pretty abusive to the public PackWriter API, because right
now we don't want to expose the real internal low-level methods
actually required to do this.
Change-Id: I437a17ceb98708b5603a2061126eb251e82f4ed4
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
DeltaIndex is a simple pack style delta generator. The function works
by creating a compact index of a source buffer's blocks, and then
walking a sliding window along a desired result buffer, searching for
the window in the index. When a match is found, the window is
stretched to the longest possible length that is common with the
source buffer, and a copy instruction is created.
Rabin's polynomial hash function is used to compute the hash for a
block, permitting efficient sliding of the window in single byte
increments. The update function to slide one byte originated from
David Mazieres' work in LBFS, and our implementation of the update
step was certainly inspired by the initial work Geert Bosch proposed
for C Git in http://marc.info/?l=git&m=114565424620771&w=2.
To ensure the encoder runs in linear time with respect to the size of
the two input buffers (source and result), the maximum number of
blocks that can share the same position in the index's hashtable is
capped at a constant number. This prevents bad inputs from causing
the encoder to run in quadratic time, but comes with a penalty of
creating a longer delta due to fewer considered copy positions.
Strange hackery is used to cap the amount of memory used by the index
to be no more than 12 bytes for every 16 bytes of source buffer, no
matter what the JVM per-object overhead is. This permits an index to
always be no larger than 1.75x the source buffer length, which is an
important feature to support large windows of candidates to match
against while packing. Here the strange hackery is nothing more than
a manually managed chained hashtable, where pointers are array indexes
into storage arrays rather than object references.
Computation of the hash function for a single fixed sized block is
done through an unrolled loop, where the first 4 iterations have been
manually reduced down to eliminate unnecessary instructions. The
pattern is derived from ObjectId.equals(byte[], int, byte[], int),
where we have unrolled the loop required to compare two 20 byte
arrays. Hours of testing with the Sun 1.6 JRE concluded that the
non-obvious "foo[idx + 1]" style of reference is faster than
"foo[idx++]", and so that is what we use here during hashing.
Change-Id: If9fb2a1524361bc701405920560d8ae752221768
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Its useful to know what the flags are or what the base that was
selected is. Dump these out as part of the object's toString.
Change-Id: I8810067fb8337b08b4fcafd5f9ea3e1e31ca6726
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
A subclass may want to use this method to release handles that are
caching reuse information. Make it protected so they can override
it and update themselves.
Change-Id: I2277a56ad28560d2d2d97961cbc74bc7405a70d4
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Searching for reuse candidates should be fast compared to actually
doing delta compression. So pull the progress monitor out of this
phase and rename it back to identify the compressing objects state.
Change-Id: I5eb80919f21c1251e0e3420ff7774126f1f79b27
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Long ago when PackWriter is first written we thought that the delta
depth could be updated automatically. But its never used. Instead
make this a simple standard setter so the caller can more directly
set the delta depth of this object. This permits us to configure a
depth that takes into account more than just the depth of another
object in this same pack.
Change-Id: I1d71b74f2edd7029b8743a2c13b591098ce8cc8f
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
C Git's fast-import uses this to determine the maximum file size
that it tries to delta compress, anything equal to or above this
setting is stored with as a whole object with simple deflate.
Define the configuration so we can use it later.
Change-Id: Iea46e787d019a1b6c51135cc73d7688a02e207f5
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
This flag will later control whether or not PackWriter search for a
delta base for this object. Edge objects will never get searched,
as the writer won't be outputting them, so they should always have
this flag set on. Sometime in the future this flag should also be
set for file blobs on file paths that have the "-delta" gitattribute
set in the repository's attributes file.
Change-Id: I6e518e1a6996c8ce00b523727f1b605e400e82c6
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
We now at least import other pack settings like pack.window, which
means we can later use these to control how we search for deltas.
The compression level was fixed to use pack.compression rather than
the loose object core.compression setting.
Change-Id: I72ff6d481c936153ceb6a9e485fa731faf075a9a
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
C Git allows this and it is quite handy.
Change-Id: I1d0238b43fca931ad2079649fb7b431e2815c351
Signed-off-by: Robin Rosenberg <robin.rosenberg@dewire.com>
We need to remember these so we can later cluster objects that
have similar file paths near each other as we search for deltas
between them.
Change-Id: I52cb1e4ca15c9c267a2dbf51dd0d795f885f4cf8
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
PackWriter wants to categorize objects that are similar in path name,
so blobs that are probably from the same file (or same sort of file)
can be delta compressed against each other. Avoid converting into
a string by performing the hashing directly against the path buffer
in the tree iterator.
We only hash the last 16 bytes of the path, and we try avoid any
spaces, as we want the suffix of a file such as ".java" to be more
important than the directory it is in, like "src".
Change-Id: I31770ee711526306769a6f534afb19f937e0ba85
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
This is an informational function used by PackWriter to help it
better organize objects for delta compression. Storage systems
can implement it to provide up more detailed size information,
or they can simply rely on the default behavior that uses the
ObjectLoader obtained from open.
For local file storage, we can obtain this information faster
through specialized routines that parse a pack object header.
Change-Id: I13a09b4effb71ea5151b51547f7d091564531e58
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
If the heap limit was set to something smaller than 8 KiB, we were
still allocating the full 8 KiB block size, and accepting up to
the amount we allocated by. Instead actually put a hard cap on
the limit.
Change-Id: Id1da26fde2102e76510b1da4ede8493928a981cc
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Added support for converting DiffEntrys to FileHeaders. FileHeaders
are DiffEntrys with a buffer containing the diff output as well as
a list of HunkHeaders. The HunkHeaders contain EditLists. The
createFileHeader(DiffEntry) method in DiffFormatter performs a Myers
Diff on the files refered to by the DiffEntry, then puts the returned
EditList into a single HunkHeader, which is then put into the
FileHeader to be returned. It also generates the appropriate diff
header an puts it into the FileHeader's buffer. The rest of the diff
output, which would normally be parsed to generate the HunkHeaders,
is not generated. In fact, the purpose of this method is to avoid
the costly diff output generation and parsing normally required to
create a FileHeader.
Change-Id: I7d8b18c0f6c85e3d02ad58995d3d231e69af5887
ReadTreeTest was hardcoded to test WorkDirCheckout. Since we want
alternative checkout implementations (especially DirCacheCheckout)
this class has been refactored so that the tests can be reused
to test other implementations
The following changes have been done:
- abstract methods for checkout and prescanTwoTrees have been
introduced. Parameters are only the two trees. As index we
will implicitly use the current index of the repo.
- whenever tests needed a manipulated index before checkout
and prescanTwoTrees it was ensured that the correct index was
persisted (before we could use not-persisted instantiations of GitIndex
passed as parameters to checkout, prescanTwoTrees
- abstract methods for getting updated, conflicting, removed entries
resulting from the last checkout, prescanTwoTrees have been introduced
- an implementation for all these abstract methods using WorkDirCheckout
has been added
- method to assert a certain state of the index and the working tree has
been added
Signed-off-by: Christian Halstrick <christian.halstrick@sap.com>
Change-Id: Icf177cf8043487169a32ddd72b6f8f9246a433f7
There were some small errors which made it
difficult to read the JavaDoc.
Change-Id: Ib3b34353465162adebaca3514d596d0edf5aea51
Signed-off-by: Stefan Lay <stefan.lay@sap.com>
The special value 127 here means how many bytes we can put into
a single insert command. Rather than use the magical value 127,
lets name it to better document the code.
Change-Id: I5a326f4380f6ac87987fa833e9477700e984a88e
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Although all modern delta decoders can process copy instructions
with a count as large as 0xffffff (~15.9 MiB), pack version 2 streams
are only supposed to use delta copy instructions up to 64 KiB.
Rewrite our copy instruction encode loop to use the lower 64 KiB
limit, even though modern decoders would support longer copies.
To improve encoding performance we now try to encode up to four full
copy commands in our buffer before we flush it to the stream, but
we don't try to implement full buffering here. We are just trying
to amortize the virtual method call to the destination stream when
we have to do a large copy.
Change-Id: I9410a16e6912faa83180a9788dc05f11e33fabae
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
We want to get rid of these APIs, because they don't perform as well
as DirCache/TreeWalk, or don't offer nearly as many features.
Bug: 319145
Change-Id: I2b28f9cddc36482e1ad42d53e86e9d6461ba3bfc
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
The encode loop had the wrong condition, objects that are 128 bytes
in size need to have their length encoded as two bytes, not one.
Change-Id: I3bef85f2b774871ba6104042b341749eb8e7595c
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Rename the ByteWindow's inflate() method to setInput. We have
completely refactored the purpose of this method to be feeding part
(or all) of the window as input to the Inflater, and the actual
inflate activity happens in the caller.
Change-Id: Ie93a5bae0e9e637b5e822d56993ce6b562c6ad15
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
We need to validate the stream state after the InflaterInputStream
thinks the stream is done. Git expects a higher level of service from
the Inflater than the InflaterInputStream usually gives, we need to
ensure the embedded CRC is valid, and that there isn't trailing
garbage at the end of the file.
Change-Id: I1c9642a82dbd76b69e607dceccf8b85dc869a3c1
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
NB.decodeInt32(info, base + 4) already returns nanoseconds.
Therefore it must not be divided by 1000000.
Change-Id: Ie8f5c4a03f984d98935dccedc2b1ba4457094899
Signed-off-by: Stefan Lay <stefan.lay@sap.com>
The FollowFilter can be installed on a RevWalk to cause the path
to be updated through rename detection when the affected file is
found to be added to the project.
The filter works reasonably well, for example we can follow the
history of the fsck command in git-core:
$ jgit log --name-status --follow builtin/fsck.c | grep ^R
R100 builtin-fsck.c builtin/fsck.c
R099 fsck.c builtin-fsck.c
R099 fsck-objects.c fsck.c
R099 fsck-cache.c fsck-objects.c
Change-Id: I4017bcfd150126aa342fdd423a688493ca660a1f
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
This way we don't have to reparse for the rename limit every time
we create a new rename detector for a repository.
Change-Id: I669d031690b85ef4da5e39189be7173fb773fc56
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Similar to what we did with diff, implement whitespace ignore options
for log too. This requires us to define some means of creating any
RawText object type at will inside of DiffFormatter, so we define a
new factory interface to construct RawText instances on demand.
Unfortunately we have to copy the entire block of common options.
args4j only processes the options/arguments on the one command class
and Java doesn't support multiple inheritance.
Change-Id: Ia16cd3a11b850fffae9fbe7b721d7e43f1d0e8a5
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Instead of crashing, output a submodule link with the simple
"Subproject commit $fullid\n" syntax used by C Git.
Change-Id: Iae8646941683fb19b73fb038217d2e3bf5f77fa9
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Passing around the OutputStream and the Repository is crazy. Instead
put the stream in the constructor, since this formatter exists only to
output to the stream, and put the repository as a member variable that
can be optionally set.
Change-Id: I2bad012fee7f40dc1346700ebd19f1e048982878
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Implement rename detection in the command line diff and log commands.
Also support --name-status, -p and -U flags, as these can be quite
useful to view more detail.
All of the Git patch file formatting code is now moved over to the
DiffFormatter class. This permits us to reuse it in any context,
including inside of IDEs.
Change-Id: I687ccba34e18105a07e0a439d2181c323209d96c
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Content similarity based rename detection is performed only after
a linear time detection is performed using exact content match on
the ObjectIds. Any names which were paired up during that exact
match phase are excluded from the inexact similarity based rename,
which reduces the space that must be considered.
During rename detection two entries cannot be marked as a rename
if they are different types of files. This prevents a symlink from
being renamed to a regular file, even if their blob content appears
to be similar, or is identical.
Efficiently comparing two files is performed by building up two
hash indexes and hashing lines or short blocks from each file,
counting the number of bytes that each line or block represents.
Instead of using a standard java.util.HashMap, we use a custom
open hashing scheme similiar to what we use in ObjecIdSubclassMap.
This permits us to have a very light-weight hash, with very little
memory overhead per cell stored.
As we only need two ints per record in the map (line/block key and
number of bytes), we collapse them into a single long inside of
a long array, making very efficient use of available memory when
we create the index table. We only need object headers for the
index structure itself, and the index table, but not per-cell.
This offers a massive space savings over using java.util.HashMap.
The score calculation is done by approximating how many bytes are
the same between the two inputs (which for a delta would be how much
is copied from the base into the result). The score is derived by
dividing the approximate number of bytes in common into the length
of the larger of the two input files.
Right now the SimilarityIndex table should average about 1/2 full,
which means we waste about 50% of our memory on empty entries
after we are done indexing a file and sort the table's contents.
If memory becomes an issue we could discard the table and copy all
records over to a new array that is properly sized.
Building the index requires O(M + N log N) time, where M is the
size of the input file in bytes, and N is the number of unique
lines/blocks in the file. The N log N time constraint comes
from the sort of the index table that is necessary to perform
linear time matching against another SimilarityIndex created for
a different file.
To actually perform the rename detection, a SxD matrix is created,
placing the sources (aka deletions) along one dimension and the
destinations (aka additions) along the other. A simple O(S x D)
loop examines every cell in this matrix.
A SimilarityIndex is built along the row and reused for each
column compare along that row, avoiding the costly index rebuild
at the row level. A future improvement would be to load a smaller
square matrix into SimilarityIndexes and process everything in that
sub-matrix before discarding the column dimension and moving down
to the next sub-matrix block along that same grid of rows.
An optional ProgressMonitor is permitted to be passed in, allowing
applications to see the progress of the detector as it works through
the matrix cells. This provides some indication of current status
for very long running renames.
The default line/block hash function used by the SimilarityIndex
may not be optimal, and may produce too many collisions. It is
borrowed from RawText's hash, which is used to quickly skip out of
a longer equality test if two lines have different hash functions.
We may need to refine this hash in the future, in order to minimize
the number of collisions we get on common source files.
Based on a handful of test commits in JGit (especially my own
recent rename repository refactoring series), this rename detector
produces output that is very close to C Git. The content similarity
scores are sometimes off by 1%, which is most probably caused by
our SimilarityIndex type using a different hash function than C
Git uses when it computes the delta size between any two objects
in the rename matrix.
Bug: 318504
Change-Id: I11dff969e8a2e4cf252636d857d2113053bdd9dc
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Alex pointed out that my description of a bare repository might be
confusing for some readers. Reword the description of the error,
and make it consistent throughout the Repository class's API.
Change-Id: I87929ddd3005f578a7022f363270952d1f7f8664
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
During code review, Alex raised a few comments about commit
532421d989 ("Refactor repository construction to builder class").
Due to the size of the related series we aren't going to go back
and rebase in something this minor, so resolve them as a follow-up
commit instead.
Change-Id: Ied52f7a8f7252743353c58d20bfc3ec498933e00
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Stefan Lay