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
By default we limit packages to publishing at-most 1000 package versions.
This aims to:
Keep the size of the version listing API end-point reasonable (which would otherwise slow down pub get).
Mitigate abuse of the system.
However, if a package hits this limit we should have an allow-list that a package can be added to such that the limit can be increased for a specific package.
We would ask package authors to reduce the publication frequency, when they are allow-listed to published more than 1000 package versions.
Ideally, it shouldn't be necessary to publish updates more than once per week (if even that much). Obviously, there can be extenuating circumstances, if there is a serious bug, security vulnerability, etc. or perhaps you're making a big release and publishing a few quick fixes.
But in general, we don't believe that it's a good idea to publish tiny changes extremely frequently. This causes churn for all the people who depend on your package, as they will be prompted to upgrade.
If on average a package is published no more than once per week (which is extremely frequent), then back of the envelop math (1000 / 52 = 19.2) suggest that 1000 versions will be enough for 19 years of package publishing.
We should probably add a section in our documentation with guidelines around publishing frequency.
The text was updated successfully, but these errors were encountered:
By default we limit packages to publishing at-most 1000 package versions.
This aims to:
pub get
).However, if a package hits this limit we should have an allow-list that a package can be added to such that the limit can be increased for a specific package.
We would ask package authors to reduce the publication frequency, when they are allow-listed to published more than 1000 package versions.
Ideally, it shouldn't be necessary to publish updates more than once per week (if even that much). Obviously, there can be extenuating circumstances, if there is a serious bug, security vulnerability, etc. or perhaps you're making a big release and publishing a few quick fixes.
But in general, we don't believe that it's a good idea to publish tiny changes extremely frequently. This causes churn for all the people who depend on your package, as they will be prompted to upgrade.
If on average a package is published no more than once per week (which is extremely frequent), then back of the envelop math (
1000 / 52 = 19.2
) suggest that 1000 versions will be enough for 19 years of package publishing.We should probably add a section in our documentation with guidelines around publishing frequency.
The text was updated successfully, but these errors were encountered: