ebe35dbefe
core: Ensure we delete duplicate files in a transaction
...
If multiple files have the same hash, we need to ensure we're not
overwriting other tempfiles in the same transaction. Instead
just delete them, since we know they're in the repo.
2012-01-15 12:25:52 -05:00
9910c0ddf4
core: Create a directory for each transaction
...
This helps ensure that we can more easily identify stale tempfiles
left over if we crash in the middle of a transaction.
2012-01-15 11:03:01 -05:00
40226c2769
core: Move triggers into separate binary: ostree-run-triggers
...
I'm trying to keep ostree as being closer to just being the versioning
filesystem, so let's split out the triggers into a different binary
(although still namespaced ostree-).
2012-01-13 15:08:27 -05:00
62821024b9
core: While walking mtrees during commit, ensure the referenced path is resolved
...
This gives us a clean error instead of an assertion.
2012-01-11 13:42:59 -05:00
5119cbdb7c
core: Add --no-xattrs option for commit
...
Will be used by ostbuild.
2012-01-05 19:10:26 -05:00
3e59cc1305
core: Add --tar-autocreate-parents option for commit
...
The tar files we're making of artifacts don't include parent
directories. Now we could change the builder to make them, but we can
also just autocreate them on import. Mode 0755 with no xattrs seems
OK here.
2011-12-23 06:49:04 -05:00
4bbc99fb1d
core: Fix build with libarchive disabled
2011-12-22 19:52:30 -05:00
dab4611263
core: Add --skip-if-unchanged option for commit
...
There's not much point for OS builds to have "empty" commits.
2011-12-22 18:47:30 -05:00
3292dcb6d6
core: Allow / in ref names like git does
...
Using / is nice basically.
2011-12-22 15:59:47 -05:00
9bcd4ff8aa
core: Use global random object
...
We were pointlessly creating a new rand (and thus opening /dev/random)
on every temp file creation, but still using the global rand.
2011-12-22 11:23:16 -05:00
d25f1bf73d
core: Use GFile for repo constructor API, and a bit more internally
...
Also, ensure that the repo directory GFile is absolute - this avoids
a getcwd() syscall every time we construct a GFile object.
2011-12-22 11:04:08 -05:00
e9fd921afe
core: Change compose to operate purely in-memory
...
This is *significantly* faster than checking out each branch into the
real filesystem, then importing it again.
2011-12-21 21:42:13 -05:00
add55849ab
core: Allow specifying a previous commit tree as a basis for a new commit
...
It's pretty trivial to map a previously existing commit tree into a
mutable tree too. While we're here change the command line arguments
for commit so that we can now properly overlay any combination of
directory, commit, or tarfile.
2011-12-21 20:10:10 -05:00
c8377c7c5c
core: Expose more details of commit via API
...
Rather than offering high level "commit directory", instead perform
operations on a mtree. Commits are treated more like regular objects.
Change the commit builtin to drive this all at a lower level.
2011-12-21 18:58:44 -05:00
59281d5901
core: Add omitted mutable tree files
2011-12-21 18:58:27 -05:00
72257bab2b
core: Rename "store" public APIs to "stage" since that's what they do
2011-12-21 17:13:04 -05:00
06548f137e
core: Make OstreeRepoFile "public"
...
* "ls" used it
* It's not like our internal API is 100% stable anyways
* The _ is ugly
2011-12-21 13:41:38 -05:00
2b7a83d1cc
core: Add new "mutable tree" class, use it for tar file import
...
The tar import code forced the resuscitation of a hackish "FileTree"
data type for representing an in-memory tree. Split this out
into an OstreeMutableTree class for future use by any other in-memory
tree construction.
2011-12-21 13:34:10 -05:00
14d682305b
core: Remove dead code from OstreeRepoFile
...
Originally I thought it would make sense for OstreeRepoFile to be
mutable, and it probably does, but we should do it in a different way.
2011-12-21 11:01:41 -05:00
bdfde03b61
core: Support committing multiple tarballs in the same transaction
...
ostbuild will generate two artifacts: foo-runtime.tar.gz and
foo-devel.tar.gz in the general case. When committing to the devel
tree, it'd be lame (i.e. slower and not atomic) to have to commit
twice.
2011-12-20 17:21:15 -05:00
11d57d63e3
core: Support --owner-uid and --owner-gid options for commit
...
This allows us to more easily import user-built tarballs into
a root-owned OSTree repo.
2011-12-18 17:36:46 -05:00
134283afbf
core: In user mode checkouts of archives, use hard links
...
Now that we've done all the gyrations to separate content from
metadata in archives, we can just hard link when doing user checkouts.
2011-12-15 16:12:31 -05:00
e014b2d9c2
core: Fix a few memory leaks
2011-12-15 15:17:27 -05:00
9a71ab187d
core: INCOMPATIBLE CHANGE: Split archive files in two parts (meta and content)
...
This will allow us to have hardlink checkouts of archives. A key use
case here is an archive repo of an OS (with root-owned files etc.)
where we want to do builds in a user tree.
A positive side effect of doing things this way is that now the SHA256
checksums for a given file should be identical regardless of whether
it's stored in an archive or bare repository.
2011-12-15 13:14:06 -05:00
ecbffd4915
core: Don't insert null byte in temporary file names
2011-12-15 13:10:41 -05:00
8ab2296a70
core: Add utility functions to convert a OstreeObjectType to/from a string
2011-12-15 13:09:51 -05:00
b8e8b58585
core: INCOMPATIBLE CHANGE: Name repo files with their type (e.g. .dirmeta)
...
This makes inspection easier. Internally the code gets simpler because
metadata and files are more unified; there is just one object type.
2011-12-12 12:29:09 -05:00
dc4164993b
core: Initialize variables in fsck
...
Otherwise we segfault on an unhandled filename.
2011-12-08 18:23:25 -05:00
f6117de8d3
core: s/pack/archived/
...
This completes the rename from the previous commit.
2011-12-08 17:45:37 -05:00
d6ceb91c35
core: Make ot_transfer_out_value() take a & for the second argument
...
As Ray Strode argued, it's confusing to have something that looks like
a function be magical. And OT_TRANSFER_OUT_VALUE is uglier.
2011-12-08 17:31:45 -05:00
35fa68a74c
core: INCOMPATIBLE CHANGE: Rename .packfile to .archive
...
It's too confusing that we call the mode "archive" but the actual
files ".packfile". Also, git already has a "packfile" that serves a
totally different purpose.
2011-12-08 17:15:59 -05:00
30c53a967c
core: Switch is_archive to an enumeration
...
This is in preparation for adding a third mode.
2011-12-08 17:05:16 -05:00
7932811fbb
core: Delete some dead code
2011-12-08 14:55:41 -05:00
dccd106731
core: Clean up checkout code
...
Take a GFile * for destination. Also, we only need one recursive
function, not two.
2011-12-08 10:28:44 -05:00
d70950e891
core: Add -U option for checkout to skip lchown() and xattrs
...
This is useful for checking out an OS filesystem tree as a regular
user, and not caring about /dev and stuff.
2011-12-06 21:38:04 -05:00
a94e0b8bf8
core: Ensure we chown created directories too
2011-12-06 20:16:03 -05:00
c32141c2bf
core: Remove some useless wrapping of packfile unpacking in checkout
2011-12-06 20:09:09 -05:00
ce7345d51e
core: Major cleanup to internal import API
2011-12-05 14:21:18 -05:00
b8000c00d5
core: Further unify code for importing from local FS and libarchive
...
Note this change makes it so we no longer call link() from an import
filesystem tree to the repository. This is a Good Thing really; it
makes local FS commits slower, but also less prone to corruption.
2011-12-05 11:20:14 -05:00
267bf8ef4a
core: Ensure we initialize checksum in FIFO case too
2011-12-05 11:15:12 -05:00
9d39517554
core: Add _from_input variant of checksum API
2011-12-05 10:28:42 -05:00
78f435d245
core: Fix handling of hardlinks for tar commits
...
We were creating files with the wrong name. Add a test.
2011-12-04 14:38:30 -05:00
17cc772cf3
core: Support for committing from any libarchive-supported format
...
We really want the ability to take a .tar.gz and directly import
it into a repository, without creating a temporary filesystem tree.
First, doing it this way is significantly faster. Also, this allows
us to handle importing tar files with e.g. uid 0 files into packed
repositories as non-root, which is very useful for tests and builds.
2011-12-04 14:08:55 -05:00
f9315e8f82
core: Fix a memory leak in checkout
2011-12-04 14:08:55 -05:00
45addeadf0
core: Handle NULL xattrs in packfile generation, throw error on invalid mode
2011-12-04 13:12:06 -05:00
ae3a5e3c07
core: Clean up filename utility API
...
Remove more unused functions, and change pathname splitting to handle
more cases like duplicate //, and to throw an error on .. as a filename.
2011-12-04 12:12:34 -05:00
1f6d49cdb9
core: Initialize checksums for non-regular files, handle NULL xattrs
2011-12-04 12:11:29 -05:00
0003e101f6
core: New ot_transfer_out_value() macro
...
This makes out values considerably less typing.
2011-12-02 18:17:06 -05:00
3ab69101f7
core: Fix packed file creation
...
We were calculating the wrong checksum, and also mistakenly
not using packed files in archive mode for non-regular files.
2011-12-02 17:49:57 -05:00
b5ce760678
core: Don't use g_file_delete(), it calls into g_vfs_get_default()
...
For some (probably crazy) reason. Skipping this and calling
unlink() directly makes 'strace' MUCH smaller.
2011-12-02 14:32:25 -05:00
Colin Walters