Re: Proposed integrated mingw packaging guidelines [Re: Planning to start unifying native and mingw packages]

On Wed, Mar 01, 2023 at 01:45:11PM +0000, Daniel P. Berrangé wrote: > Our goal is to strongly encourage the use of integrated mingw packaging, > but still allow native package maintainers the discretion to opt-out of > this if they feel strongly against handling mingw themselves. The keys > terms of the updated guidelines are this paragraph: > > [quote] > * Where the same Fedora contributors intend to maintain both the native > and MinGW builds of a package, they **MUST** use the integrated packaging > approach. > > * Where the upstream project supports the Windows platform as an official > build target and has automated CI, contributors **SHOULD** prefer the > integrated MinGW packaging approach. While native package maintainers are > encouraged to accept this, they may decline this suggestion at their > discretion. When I first read this, it seemed like it was saying that if the upstream project supports Windows we (strongly) should produce a mingw package. I understand this is not what you're saying, but perhaps it could be clarified by adding "... and the Fedora packager wishes to provide a mingw library"

> * Where the upstream project does not have automated testing of Windows > builds, the MinGW package support **MAY** use the integrated packaging > approach, subject to agreement of the native package maintainer. > > * Where the upstream project only supports Windows builds, the separate > packaging approach **MUST** be used. There will be no corresponding > native package in Fedora expected. This situation is rare. I'm actually trying to think of a case. Maybe the tools I wrote like mingw-nsiswrapper and mingw-crossreport? The second one should probably be removed, since it's not very relevant today.

> * When a contributor proposes a new native package to Fedora that provides > libraries that are known to support Windows, the reviewer **SHOULD** > inquire whether the contributor would like to add MinGW builds at the > same time. The contributor **MAY** decline this suggestion at their > discretion. > [/quote]