We could have a full ICS calendar:
- Generate all releases
- Update the ICS file to have better metadata for subscription
- Study the load impact on the server? Can we indicate a refresh time to clients to not hammer the server?
- Add some file caching to the generation?
- Test with a few popular web and desktop clients?
The benefit would be that any release schedule change would be updated automatically.
A drawback compared to a raw import is that all of our logic is based on the Release class which only deals with upcoming releases. Pre-release milestones would disappear as soon as we ship. We could need to have some logic to at least generate data for the current release. Probably also repopulate past releases from product-details.
The current ICS also does not have a lot of tests as relman is the only consumer when we import a release to gcal.
We could have a full ICS calendar:
The benefit would be that any release schedule change would be updated automatically.
A drawback compared to a raw import is that all of our logic is based on the Release class which only deals with upcoming releases. Pre-release milestones would disappear as soon as we ship. We could need to have some logic to at least generate data for the current release. Probably also repopulate past releases from product-details.
The current ICS also does not have a lot of tests as relman is the only consumer when we import a release to gcal.