You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The config/gpi_pipeline_primitives.xml file has to be auto-generated by the "Rescan Pipeline Config" whenever new primitives are added. Since it's a machine-generated data file it probably shouldn't be in the version control system. If it is, then it adds more work to frequently commit changes to that which don't really need to be tracked since it's derived from primitives metadata.
But if we take it out, that means that everyone is going to have to remember to rescan their pipeline configs when they pull in new changes, and new users are going to need to do that after installing the pipeline before they can run it. So that adds extra steps too.
On Oct 12, 2016, at 6:25 AM, Marshall Perrin [email protected] wrote:
The config/gpi_pipeline_primitives.xml file has to be auto-generated by the "Rescan Pipeline Config" whenever new primitives are added. Since it's a machine-generated data file it probably shouldn't be in the version control system. If it is, then it adds more work to frequently commit changes to that which don't really need to be tracked since it's derived from primitives metadata.
But if we take it out, that means that everyone is going to have to remember to rescan their pipeline configs when they pull in new changes, and new users are going to need to do that after installing the pipeline before they can run it. So that adds extra steps too.
Correct me if I'm wrong, but I believe the hooks are not on version control so we'd something like a script that writes the appropriate git hook into each person's .git directory.
On launch of the pipeline, can we check the date modified of gpi_pipeline_primitives.xml against the date modified of all files in the primitives directory? If any of the primitives have been modified, then we trigger a rescan of the pipeline config on startup of the pipeline.
The
config/gpi_pipeline_primitives.xml
file has to be auto-generated by the "Rescan Pipeline Config" whenever new primitives are added. Since it's a machine-generated data file it probably shouldn't be in the version control system. If it is, then it adds more work to frequently commit changes to that which don't really need to be tracked since it's derived from primitives metadata.But if we take it out, that means that everyone is going to have to remember to rescan their pipeline configs when they pull in new changes, and new users are going to need to do that after installing the pipeline before they can run it. So that adds extra steps too.
Which is preferable? @semaphoreP @kfollette opinions?
The text was updated successfully, but these errors were encountered: