On 2021-08-03 12:36, Achim Gratz wrote:
Brian Inglis writes:
But I didn't believe Cygwin Setup would replace packages that are
still dependencies of installed packages, unless you explicitly ask it
to remove obsolete packages, rather than unneeded dependencies.
Unless it was due to my marking libidn12 as obsoleting libidn11?
You never do that in this situation, since it tells setup to install
libidn12 while removing libidn11, which it dutifully did. Why'd you
even get the idea that you should obsolete libidn11?
I assumed that existing package dependencies would override the
obsoletion of libidn11, rather than the obsoletion overriding.
Maintainers: is that something that we should not do on a package ABI
break?
No, not at all. You never obsolete library packages unless they really
need to be removed and replaced by something else.
How should maintainers handle such situations in cygport?
Nothing to do there, move along. the older library version continues to
exist and is used until all packages that depend on it have updated.
I would like to know the correct approach to take to mitigate this and
future such situations before I create a curl -2 package release.
What you should check is that there is no transitive dependency to
libidn11 left in the dependency chain for anything that depends on a
package that uses the newer version. If that is possible you ideally
need to update all involved packages together.
Thanks, now I know not to do that.
Is Jon's change all that is required, or do I need to build and upload a
libidn11 to replace the obsoleted version, as well as a current libidn12
release with the obsolete removed?
--
Take care. Thanks, Brian Inglis, Calgary, Alberta, Canada
This email may be disturbing to some readers as it contains
too much technical detail. Reader discretion is advised.
[Data in binary units and prefixes, physical quantities in SI.]
--
Problem reports: https://cygwin.com/problems.html
FAQ: https://cygwin.com/faq/
Documentation: https://cygwin.com/docs.html
Unsubscribe info: https://cygwin.com/ml/#unsubscribe-simple