The biggest difference is that without binaries it becomes possible again to checkout dist again and clean it up without solely relying on svn import/export. That is way more comfortable and allows to use other tools than CLI for that.
Regarding filtering what needs to be put to dist, that can be easily scripted (while the cleanup unfortunately can't be easily scripted). Konrad > On 25. Feb 2021, at 12:12, Carsten Ziegeler <[email protected]> wrote: > > You should have automation in place to detect issues - that used to be in > place, not sure if it still is. As soon as its more than 10 files any manual > process will fail. > > If we start to exclude files from being put into dist, I think the release > process is even harder than today - as I need to do the filtering and > remember what to not put there. Today it's easy, get everything from nexus > and put them up there > > Regards > Carsten > > > Am 25.02.2021 um 09:54 schrieb Konrad Windszus: >>> What exact problems are you seeing? >> https://issues.apache.org/jira/browse/SLING-10153 >> https://issues.apache.org/jira/browse/SLING-10152 >> https://issues.apache.org/jira/browse/SLING-10151 >> In general I doubt that dist is very often used for downloading binaries and >> I want to reduce effort for doing releases. >> It is close to impossible to detect issues like these with 6k files without >> tooling.... >> Konrad > > -- > -- > Carsten Ziegeler > Adobe Research Switzerland > [email protected]
