| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
|/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In order to make releases and freeze system branches entirely, we
need to be able to 100% petrify a system branch (that is, resolve
ALL refs into SHA1s) and tag this state to be able to check it out
again later.
This is essentially what "morph tag" does. It takes a tag name
and an arbitrary amount of arguments to "git tag", petrifies all
morphologies of the current system branch behind the scenes, creates
a dangling commit and attaches an annotated tag to it.
Petrifying in this case means that all refs used for chunks are
resolved into commit SHA1s. For stratum and system morphologies,
the refs are replaced by the name of the tag that's being created.
The "tag" command also supports tagging when stratum morphologies
are spread across multiple repositories. In this case, it will
include all statum morphologies from other repos in the tag commi
in the branch root repo. The references to these morphologies are
updated so that they point to the branch root repo and the tag
being created.
This commit also adds a few tests for "morph tag" to verify that
all this works.
|
|
|
|
| |
Be consistent about placement of test description, blank lines, etc.
|
|
|
|
|
|
|
|
| |
Most of the bespoke logic for the version check is unnecessary,
since the output to display can be easily inferred from the filename.
This fixes some test failures where the version check would cat a
file to fake the output, but fail because the file was removed.
|
|\
| |
| |
| | |
git://git.baserock.org/baserock/morph
|
| |
| |
| |
| |
| | |
Currently the message is still displayed "Updating xxx" but no
update is actually done.
|
|\ \
| |/
| |
| | |
git://git.baserock.org/baserock/morph
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Some tests already used test: and in order to be able to share the
scripts/setup-3rd-party-strata script they need to all use the same
prefix.
Using baserock: implies that we are using real code from Baserock,
so I picked test: because the tests only ever use mock morphologies
and no real code.
|
|/ |
|
|
|
|
| |
Merge is by far the most complex of the branching and merging commands.
|
|
|
|
|
|
|
|
|
|
| |
The output of the list-tree depended on the value of TMPDIR.
Instead we replace this with a test that the required directory exists.
merge-with-stratum-renamed also fails with a TMPDIR different to /tmp,
but I have no idea why.
Signed off by Daniel Silverstone in person
|
|\
| |
| |
| | |
Reviewed-By: Richard Maw <richard.maw@codethink.co.uk>
|
| |
| |
| |
| |
| | |
Output needs to be stable, not least so that the test doesn't fail
sporadically.
|
|/
|
|
|
|
|
|
|
|
|
|
|
| |
This change causes 'morph petrify' to avoid petrifying any chunk whose
ref matches the current system branch, because it makes no sense to
petrify something that is also being edited. It also improves efficiency
slightly and adds warning where different systems point to different
refs of the same stratum.
A non-obvious effect of this is that if you try to petrify 'master',
many of the chunks won't get petrified because they are built from
'master'. However, petrifying master makes no sense so I'm not sure
that we need to worry.
|
|
|
|
| |
This provides a user-friendly summary of the workspace or branch status.
|
|\
| |
| |
| |
| |
| | |
'origin/samthursfield/S4873-warn-when-merge-causes-petrification'
Renamed petrification test slightly as merge fixup.
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The concept of a component path is new. This is simply a concise way
of referring to a component in an error message, and looks like this:
base-system-x86_64-generic.bsp-x86_64-generic.linux
We currently only touch the 'edited chunks' in merge_stratum(), i.e.
those in the FROM branch where 'morph edit' was run. However, the
petrification can affect any chunk so there is a new method added to
obtain all components in a morphology. This function also returns the
differences between the two, which we will make use of at a later date.
|
|/ |
|
|
|
|
|
|
|
|
|
|
| |
This should not normally be used, because we make no attempt to detect
when a full URL and a keyed URL are equivalent, so they cannot be used
interchangably.
However, 'foreach' would previously fail completely if the branch root
happened to be a full URL because it didn't call convert_uri_to_path()
correctly.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We often have .gitmodules edited to contain a URI such as
upstream:gnulib, so that we can transparently mirror these in different
locations.
It would be nice to set up git url.insteadOf rules to expand these for
the submodules, but 'git submodule update' uses 'git clone' to fetch
them, which will not take into account the configuration of the
parent repository.
Instead, we set up the submodules automatically and rewrite the URLs
directly in the configuration. The user will need to recreate their
system branch checkouts if their URL configuration changes, or update
the URLs manually, but that should not happen often.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Upsides:
- clearer error messages on conflicts (we no longer dump the git
output)
- merge base is available during morphology merging
- we no longer need to go through the complexity of implementing a
git merge driver
We now manually fetch and then merge, instead of using git pull. This
is not strictly necessary, but it makes it clearer in the code how
FETCH_HEAD is involved in the process.
|
|
|
|
|
|
|
|
|
|
| |
The rationale here is that inside a checkout of a system branch, the
user should only be committing to the refs for that system branch,
because those are the only ones that 'morph merge' will look at.
This removes a problem where we could be confused by a ref with a
name that would be sorted before 'master' in the result of 'git
show-ref'.
|
|
|
|
|
|
| |
Previously we had to manually use git merge to resolve the ref
conflicts in the morphologies, but the morphology merge driver
now takes care of them.
|
|
|
|
|
|
|
|
|
|
| |
We now have a two-stage merge process. Stage one only runs if there
are changes in both branches on the same file. At the end of stage
one we assume that all the components that were edited in the FROM
branch still have their 'ref' field set to the FROM branch. Morph
then iterates through these repositories, performing a merge in each
one, and then updates the refs in the morphologies again with the
correct target branches.
|
|
|
|
|
| |
This code made no sense, but since the previous command was expected
to fail it didn't have any effect on the test.
|
|
|
|
|
| |
These do need to be quite clear, because merge failures will happen a
lot and we are relatively unhelpful at the moment.
|
| |
|
| |
|
|
|
|
|
| |
Our non-destructive morphology editing requires at least Python 2.7,
which is not available on Squeeze.
|
|
|
|
|
| |
Make sure we don't lose the 'unpetrify-ref' field in this case, or
unpetrify will break.
|
|
|
|
| |
Split out the external strata scripts to share between workflow tests.
|
|
|
|
| |
Tested in the petrify test, to avoid duplication.
|
| |
|
|
|
|
|
|
| |
Use cliapp.Application.runcmd_unchecked() instead of runcmd() to get
the full command output, without requiring AppException.msg (which is
an inferior way to get the output in any case)
|
| |
|
| |
|
|
|
|
|
| |
We now add the strata in one branch, merge, and then make the changes
in another.
|
|
|
|
|
|
|
|
|
|
|
|
| |
* Spot completely unmergable cases
* Make sure that changes in a chunk repo are merged to the back to
that chunk's original ref, which is not necessarily the same as
the TO system branch.
* Test renaming a chunk (works fine if repo is the same)
* Try adding a new chunk or stratum and editing it too.
|
|
|
|
| |
'git commit' fails if there are no changes in any case.
|
|
|
|
|
|
| |
Rework the test outputs to match reality. This is due to tweaks in how morph
edit/checkout gets its repository now that the copy-from-cache is marginally
more like a traditional clone in terms of what you get at the end.
|
|
|
|
|
|
| |
This reworks the blackbox tests to work with the bare repository caches. For
the most part it's slight changes to error messages and tweaks to ignore the
repository caches during file listing.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
This requires disabling the feature that retains the original order of
fields in a morphlogy when it gets overwritten. The implementation relies
on features that are not available in Python 2.6. We need to support
Morph on Debian squeeze, for bootstrapping purposes, and therefore need
to have it work with Python 2.6. However, the morphology rewriting is
only relevant for system branching and merging, and that isn't needed
for bootstrapping, so we disable the affected tests on Python 2.6.
|
|
|
|
|
|
| |
Move this into a script which can be sourced by the 'setup' scripts
and the actual tests (this is needed as the environment in 'setup' is
not passed on to the tests).
|
|\
| |
| |
| | |
git://roadtrain.codethink.co.uk/baserock/morph
|
| |
| |
| |
| | |
Update merge test to check the message.
|
|/
|
|
|
|
|
|
|
| |
This gives us consistency with morphologies, where the triplets are
repo|ref|morphology, not repo|ref|filename
Anyone who runs 'morph build baserock:morphs master system.morph' will
now see an error ending with 'was looking for system.morph.morph', which
should make it clear where they have gone wrong.
|
| |
|
| |
|