summaryrefslogtreecommitdiff
path: root/ybd
Commit message (Collapse)AuthorAgeFilesLines
* Explicitly check that track is TrueBen Brown2018-02-231-1/+1
|
* Allow the baserock directory to be used for rpm related filesbenbrown/rpm-baserock-dirBen Brown2018-02-231-4/+19
| | | | | Allows rpm specific files to be hidden away in the baserock directory during rpm generation.
* Log if we were unable to find the manifest to extractBen Brown2018-02-201-1/+5
| | | | | | | This needn't cause an error, if it is required the component will fail during rpm generation. If the path is relative to the metadir, skip without warning.
* Remove .build and .inst directories from deploymentbenbrown/deployment-dirsBen Brown2018-02-161-0/+5
|
* Bump artifact-versionBen Brown2018-02-151-1/+3
|
* Have a separate cache-key for rpm building with rpm-metadataBen Brown2018-02-152-17/+22
| | | | | | When changing rpm-metadata, we don't want to rebuild the artifact we use to generate the rpms, we just want to regenerate the rpm using the same artifact.
* Extract manifests from the root directory into the metadir prior to packagingBen Brown2018-02-151-1/+14
|
* Remove extra '%' from common argsBen Brown2018-02-141-4/+3
|
* Remove extra '%' in spec descriptionBen Brown2018-02-141-1/+1
|
* Bump artifact-versionbenbrown/prefix-and-bd-renamesBen Brown2018-02-131-1/+3
|
* Add prefix to cache-keyBen Brown2018-02-131-0/+3
|
* Include build-depends cache-keys as a list in hash factorsBen Brown2018-02-131-3/+9
| | | | | | | | | Previous behaviour saved cache-keys by component path. If the path of a morph were modified, but not it's contents, the cache-keys of the components reverse-depends would change. Save the cache-keys of the build dependencies as a list instead, ignoring the path.
* Fix rpm/rpmbuild error checkingBen Brown2018-02-131-2/+2
|
* Improve error message on failure to generate rpm name from metadatabenbrown/misc-rpm-fixesBen Brown2018-02-091-3/+2
|
* Improve error message on missing rpm/rpmbuild binariesBen Brown2018-02-091-2/+7
|
* Add /var to the writable paths to fix linux-user-chroot rpm buildsBen Brown2018-02-091-1/+1
|
* Log the path RPMs have been deployed toBen Brown2018-02-091-1/+1
|
* Only log copying to deployment area when running verboselyBen Brown2018-02-091-1/+1
|
* Translate additionally supported list format for rpm-metadata at parse timeBen Brown2018-02-092-14/+10
|
* Create deployment dir earlyBen Brown2018-02-091-6/+5
|
* Pass assembly.compose during rpm packagingBen Brown2018-02-091-1/+1
| | | | | To cover the case where component artifacts have been removed from the cache, while the "system" artifact remains.
* Ignore all hidden directories when looking for definitionsSam Thursfield2017-10-241-13/+17
| | | | | | | | | | | | | | | This un-does a change done in commit f072299d04 which caused YBD to stat() every directory under the current working directory when looking for definitions. This was causing issues on GitLab CI, where we're required to put the YBD cache inside our checkout of definitions.git. The cache contains 10,000s of files and calling stat() on each one is a massive slowdown. This change also causes YBD to ignore any hidden directory when searching for definitions rather than just `.git`. This allows us to put the cache at `./.cache` without issue.
* Fix config access, it is in the app moduleBen Brown2017-10-171-1/+1
|
* Add tag check to for-each-ref replacementBen Brown2017-10-161-6/+8
|
* Use branch in place of for-each-refBen Brown2017-10-131-2/+2
| | | | for-each-ref didn't get --contains until v2.7.0.
* Print which file is used should multiple files be matchedBen Brown2017-10-131-0/+1
|
* Exit should multiple file match the prebuilt rpm patternBen Brown2017-10-131-1/+2
| | | | Only applies if check-definitions is set to 'exit'.
* Sort the glob matches so order consistent across filesystemsBen Brown2017-10-131-1/+2
|
* Continue if the lockfile no longer existsbenbrown/lockfilesBen Brown2017-09-261-4/+8
|
* Only delete file if this process (thread) locked it and has finished with itBen Brown2017-09-261-1/+3
|
* flock throws IOErrorsBen Brown2017-09-261-1/+1
|
* Only write sub-package 'vendor' tags if explicitly setjjardon/vendor-in-all-packagesBen Brown2017-09-261-4/+2
| | | | Sub-packages will inherit from the top level declaration.
* Add vendor tag to all non main packagesMartin Brook2017-09-261-0/+4
|
* Exit/warn if a component ref does not contain the shaBen Brown2017-09-071-3/+36
|
* Filter .trees from artifacts for cullingBen Brown2017-09-072-2/+2
| | | | Rather than saving after cull.
* Don't reuse entry from trees if we're tracking-branchBen Brown2017-09-071-1/+2
|
* Split tracking-branch logic into its own methodBen Brown2017-09-071-2/+6
|
* Warn if unpetrify-ref is set, it is now deprecatedbenbrown/sanitise-ref-shaBen Brown2017-09-061-0/+3
|
* Sanitise track checkingBen Brown2017-09-042-9/+7
| | | | | Updated to use ref if tracking or no sha specified. unpetrify-ref is deprecated.
* Use sha in place of refBen Brown2017-09-043-4/+4
| | | | This will be set by get_tree() at this point.
* Warn if 'ref' is not a sha1, assuming 'sha' is not providedBen Brown2017-09-011-3/+8
|
* Use ref and sha in manifestBen Brown2017-09-011-4/+2
| | | | unpetrify-ref use is deprecated.
* Remove magic ref/sha logicBen Brown2017-09-013-15/+1
| | | | | This has caused far more trouble than it's worth, just be explicit about what should be used.
* Return (tree, sha) pair with a single callBen Brown2017-09-011-5/+3
|
* Save tree entry for reuseBen Brown2017-09-011-4/+5
|
* Write cache-key to results file every timeBen Brown2017-09-011-1/+1
| | | | | This is useful for rpm generation, to determine the path rpms are deployed to programmatically.
* Return early if .trees file does not existBen Brown2017-09-011-0/+2
|
* Remove erroneous double quotebenbrown/rpm-deploy-no-sandboxBen Brown2017-08-301-1/+1
|
* Stop using the sandbox to query rpm metadataBen Brown2017-08-301-28/+13
| | | | | | | At this point, we just need to read any metadata from the rpm, this doesn't require the sandboxed environment. This should considerably speed up rpm deployment time.
* Use 'gigabytes' instead of magic numberps-kbas-cull-v2Paul Sherwood2017-08-291-1/+2
|
View Lorry Controller Status