Bernard Normier

We should do the same in C++ with a `std::optional`. See: https://github.com/zeroc-ice/ice/blob/69a59072e496c2e13d1cfe79baf779905f731e6e/cpp/src/Ice/Reference.h#L160 and https://github.com/zeroc-ice/ice/blob/69a59072e496c2e13d1cfe79baf779905f731e6e/cpp/src/Ice/Reference.h#L61

> I don't understand how the idleTimeout tests works. > > Since back pressure works as of this PR, and the idle check is currently back-pressure unaware, the first test...

Similar failure with the idleTimeout test: ``` *** [106/259] Running csharp/Ice/idleTimeout tests *** [ running client/server test - 06/21/24 17:43:26 ] - Config: ssl,x64 (dotnet D:\a\ice\ice\csharp\test\Ice\idleTimeout\msbuild\server\net8.0\server.dll --Ice.Default.Host=127.0.0.1 --Ice.Warn.Connections=1 --Ice.Default.Protocol=ssl --Ice.IPv6=0...

I committed a temporary work-around for this failure.

A related question is: do we want to allow (or keep allowing) sharing configuration files between Ice applications written in different languages? If the answer is yes, then it makes...

Duplicate of #2746.

I am ok with changing this behavior and letting FacetNotExistException through. Note however that this is a thoroughly documented and tested behavior: https://doc.zeroc.com/ice/3.7/client-server-features/facets > If an Ice object does not...

This C# doc-comment does not match the C# implementation and does not match the documented behavior in https://doc.zeroc.com/ice/3.7/client-server-features/facets. I think we should either: a) keep the documented behavior and fix...

In theory, there is a third option: c) change the checkedCast behavior to always catch FacetNotException and return null for FacetNotException That's what we currently do in Java. Pros: like...

I am going to implement the proposed behavior - checkedCast does not handle any exception thrown by ice_isA.