Bruce - The primary gist of it is, client applications do not use the preconfigured classpath determined by Geode itself, such as would be the case when you start servers using *Gfsh*. Clients are not started with *Gfsh*, or any other Geode script for that matter.
On Thu, Feb 27, 2020 at 8:53 AM Udo Kohlmeyer <u...@apache.com> wrote: > @Bruce, Thank you for bringing this up. The first iteration of this > would specifically target only the server-side. BUT, I do agree, that > clients could suffer similar problems. Yet, from experience, I (and many > other users) have deployed different versions of the Spring Framework > (compared to Geode) with success. > > Generally it is easier to resolve client-side dependencies than server, > as one has more control over the client than a deployed (managed) server > instance. > > But if we think that there is a specific scenario that a client is > suffering from, I would be more than happy to make sure that this is > rolled out to the client shortly after having delivered it into the server. > > --Udo > > On 2/27/20 8:45 AM, Bruce Schuchardt wrote: > > Udo, how does this relate to the client cache? I assume people have the > same problems with dependencies in client-cache applications that they have > in functions that they deploy on a server-cache. > > > > On 2/26/20, 10:10 AM, "Udo Kohlmeyer" <u...@apache.com> wrote: > > > > Hi there Geode Dev's. > > > > There is a new RFC proposal on ClassLoader Isolation. The review end > > date is 13 Feb 2020. > > > > > https://cwiki.apache.org/confluence/display/GEODE/ClassLoader+Isolation > > < > https://cwiki.apache.org/confluence/display/GEODE/ClassLoader+Isolation> > > > > Please review and discuss in this thread. > > > > --Udo > > > > > > > > > -- -John Spring Data Team