Conversation
|
Cache hit rate is higher than I thought:
|
|
It does not look correct to use |
|
The MythTV buildbot network is the official CI. The github runners are just advisory as the quick and dirty testing. I am not a decider, but if you want to actually usefully use the few github runners (left with a ccache that can be supported under the github total cache limits), I would create runners that run clazy and tidy to prevent stupid PRs that only make things worse (i.e. PR's that introduce new issues will be marked as such and rejected immediately). A number of years ago I looked at such runners and found that the recipe for the running of such in the official buildbot network was not complete/sufficient for github runners (that may have been my misunderstandings, of course). |
|
@stuarta @garybuhrmaster is all of the buildbot stuff at https://github.com/MythTV/buildbot-utils? |
|
This: ansible_cmd: 'ansible-playbook --limit localhost -e "qt6=true" mythtv.yml' This would pass true as a bool: ansible-playbook --limit localhost --extra-vars "{'qt6':true}" mythtv.yml I can provide additional detail if needed. [just -e is OK] |
25f3f80 to
adaf684
Compare
|
@billmeek It now uses the correct parameter passing syntax for a Boolean. I had just copied what was already in the file. |
adaf684 to
1735a02
Compare
This will make the post run statistics show how effective the cache was for just that compile.
This will prevent caches from pull requests from evicting the caches from master.
Cache hit rate was and cache size is Linux configure: about 92.5%, now just over 250M Fedora CMake: about 80%, now 300M macOS: 100%, just under 250M
1735a02 to
902d09d
Compare
I intend to add an aarch64 Linux build and Android and MinGW cross compiled builds.
See also comments from #1253
I think I should have reopened the pull request before force pushing to it.