| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
* mkdir needs -p to not complain about already-existing directory
|
|
* the sanity-check should only run, when nothing else runs
* this enables some level of clean-up: intentions, that never successfully ran, will be tried again, here
|
|
* no need to trouble the db, if we don't run, because we got no lock
|
|
* now, it's safe to call it in almost all places
* this enables us to do it at the start of many scripts, potentially cleaning up old left-over intentions
|
|
intention needs to be re-executed
|
|
|
|
|
|
intention, if the db query fails
* the `sort -u` was unnecessary anyways
* the `sort -u` swallowed the exit code of mysql_run_query
* failing mysql_run_query should stop the execution flow, i.e. not remove the intention
|
|
* try 15 times instead of 10
* increase time interval from trial to trial, hopefully hitting an empty spot sometime
|
|
|
|
* aid in finding which tool broke, if there are residual files on the buildmaster
|
|
* get rid of the deprecation warning
|
|
* Do not lock build list, so builders can get jobs.
* This may lead to builders getting for jobs, that are currently overwritten or deleted. This should not be worse than if the builder got the job before running get-package-updates, though.
|
|
|
|
* use null-revisions, if this part is not needed, otherwise make_source_info et al. will fail
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
* they were not populated anyways
|
|
|
|
packages with a complete PKGBUILD
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|