| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|\ |
|
| |
| |
| |
| |
| | |
Making the changes mentioned in the comment would be quite invasive,
and we don't have to do so quite yet.
|
| | |
|
| | |
|
|/
|
|
|
| |
The artifact's build dependencies replace the build order graph
from previously.
|
|\ |
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Previously the code would edit strata that dependended on the stratum
being edited, but would ignore the dependency chain beyond that. In
fact, we need to edit all strata in the dependency chain to avoid
having two different versions of a stratum in the same build.
This splits the modification into two steps: changing the stratum that
points to the chunk, and recursively changing references to any strata
that have been altered.
|
|\ \
| | |
| | |
| | | |
Reviewed-By: Richard Maw (in real life)
|
|/ /
| |
| |
| |
| | |
Also, make test use bash instead of sh, so it passes on squeeze
as well as Baserock.
|
|\ \
| | |
| | |
| | | |
Reviewed-By: Richard Maw (on irc)
|
| | |
| | |
| | |
| | | |
Suggested-By: Richard Maw
|
|/ /
| |
| |
| | |
Reported-By: Richard Maw
|
|\ \
| | |
| | |
| | | |
git://git.baserock.org/baserock/baserock/morph
|
| | |
| | |
| | |
| | | |
Reported-By: Richard Maw
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This allows code sharing amongst all the places that create a
system in a raw disk image. This also adds the creation of
a factory-run subvolume, and fixes error messages for errors
that happen during a disk image creation.
Suggested-By: Richard Maw
Suggested-By: Sam Thursfield
|
| | |
| | |
| | |
| | | |
Suggested-By: Richard Maw
|
| | |
| | |
| | |
| | | |
Suggested-By: Richard Maw
|
| | |
| | |
| | |
| | | |
Suggested-By: Richard Maw
|
| | |
| | |
| | |
| | | |
Suggested-By: Richard Maw
|
| | |
| | |
| | |
| | | |
Suggested-By: Sam Thursfield
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This saves a check (and an indentation) in the deployment plugin,
making the code a tiny bit simpler.
Suggested-By: Sam Thursfield
|
| | |
| | |
| | |
| | | |
Suggested-By: Sam Thursfield
|
| | |
| | |
| | |
| | | |
Suggested-By: Sam Thursfield
|
| | |
| | |
| | |
| | |
| | |
| | | |
This is necessary so that the write extension run by cmdtest finds
the morphlib of the source tree we're running ./check in, rather
than the system installed one.
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This adds a new optional field to system morphologies:
"configuration-extensions".
The deployment plugin relies heavily on code from the branch and
merge plugin. This needs to be eventually fixed by refactoring
the codebase so that the shared code is in morphlib and not in
plugins. However, doing that is beyond the scope of adding a
deployment plugin.
|
|\ \ |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Newer versions of git have changed the behaviour when cloning with
--branch=$sha1. It used to be nearly synonamous with clone then checkout.
The clone then checkout logic is different to the clone with branch
logic, since they have different behaviour with regard to the branches
that are checked out in the workspace.
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The test builds a system, makes a commit to mainline, then branches
from the system and checks whether it is petrified to the exact commits
that the system was built from by checking the morphology contents and
rebuilding the system.
|
| | |
| | |
| | |
| | |
| | |
| | | |
Branch from image takes a directory containing metadata, then creates
a branch of what the System was built from and petrifies it to when
the System was built.
|
| | | |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Pass resolved_refs in with the key as a repo, ref pair to petrify
to the value, rather than petrifying to the current state of the ref.
Set update_working_tree to also change files in the working tree,
instead of just the index.
|
| | |
| | |
| | |
| | |
| | |
| | | |
Applications default to logging to console, which causes unwanted
output during tests, and I thought this was tidier than allowing
tests to override the logger per-call.
|
| | |
| | |
| | |
| | | |
This returns a sorted list of possible aliases for a url.
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This also involved codifying it ignoring malformed aliases.
Such behaviour was implemented before, but wasn't obvious.
The restructuring made it have a line of code of its own,
which meant it needed unit testing.
|
| | | |
|
|/ /
| |
| |
| |
| |
| |
| |
| |
| | |
This is required for reproducing the morphologies that created the
artifacts.
It is possible, but expensive, to `unexpand` a url, provided your
repository aliases are the same as the morphology's, but this may
not always be the case.
|
|\ \ |
|
| | |
| | |
| | |
| | | |
No longer possible for this to happen.
|
| |/
| |
| |
| |
| |
| |
| | |
Allowing differences causes only confusion. Whether or not the 'name'
field should surive at all is up for debate: it's very useful when
seeing the contents of the morphology out of context (e.g. in
documentation or pastebins), but is that worth the extra effort?
|
|\ \
| |/
|/| |
|
|/
|
|
|
| |
This is equivalent to using yaml.safe_load(), but lets us customise
how yaml is parsed.
|