Package Refresh hangs when used Packages are from a repository that uses Global Protect VPN without being connected

This report highlights the usability issues in the case of a service being unreachable that does not fail with an error immediately but instead times out eventually (depending on the default OS timeout). This situation can lead to essentially locking up the Editor as soon as there is a non-trivial number of packages in the project. Some VPN-protected servers would treat requests from clients not connected to the VPN if it didn't exist, simply not responding to the connection request. Other VPN configurations might instead immediately refuse the connection or respond with an HTTP error response (such as 403), resulting in immediate failure instead of potentially very long timeouts. We understand that saying this doesn't help, but we thought it might help provide a bit more context on the issue. Unfortunately, there is no single way for us to fix this problem. We do have several ideas on how to improve this so that in the given situation, such as implementing faster connection timeouts or specifically detecting timeouts and eagerly skipping subsequent requests to the same server, such that the Package Manager does not end up blocking the Editor for so long. We plan on implementing some of these improvements gradually, but we are not being able to provide any ETA on these improvements, for which we apologize.

This report highlights the usability issues in the case of a service being unreachable that does not fail with an error immediately but instead times out eventually (depending on the default OS timeout). This situation can lead to essentially locking up the Editor as soon as there is a non-trivial number of packages in the project. Some VPN-protected servers would treat requests from clients not connected to the VPN if it didn't exist, simply not responding to the connection request. Other VPN configurations might instead immediately refuse the connection or respond with an HTTP error response (such as 403), resulting in immediate failure instead of potentially very long timeouts. We understand that saying this doesn't help, but we thought it might help provide a bit more context on the issue. Unfortunately, there is no single way for us to fix this problem. We do have several ideas on how to improve this so that in the given situation, such as implementing faster connection timeouts or specifically detecting timeouts and eagerly skipping subsequent requests to the same server, such that the Package Manager does not end up blocking the Editor for so long. We plan on implementing some of these improvements gradually, but we are not being able to provide any ETA on these improvements, for which we apologize.

This report highlights the usability issues in the case of a service being unreachable that does not fail with an error immediately but instead times out eventually (depending on the default OS timeout). This situation can lead to essentially locking up the Editor as soon as there is a non-trivial number of packages in the project. Some VPN-protected servers would treat requests from clients not connected to the VPN if it didn't exist, simply not responding to the connection request. Other VPN configurations might instead immediately refuse the connection or respond with an HTTP error response (such as 403), resulting in immediate failure instead of potentially very long timeouts. We understand that saying this doesn't help, but we thought it might help provide a bit more context on the issue. Unfortunately, there is no single way for us to fix this problem. We do have several ideas on how to improve this so that in the given situation, such as implementing faster connection timeouts or specifically detecting timeouts and eagerly skipping subsequent requests to the same server, such that the Package Manager does not end up blocking the Editor for so long. We plan on implementing some of these improvements gradually, but we are not being able to provide any ETA on these improvements, for which we apologize.

This report highlights the usability issues in the case of a service being unreachable that does not fail with an error immediately but instead times out eventually (depending on the default OS timeout). This situation can lead to essentially locking up the Editor as soon as there is a non-trivial number of packages in the project. Some VPN-protected servers would treat requests from clients not connected to the VPN if it didn't exist, simply not responding to the connection request. Other VPN configurations might instead immediately refuse the connection or respond with an HTTP error response (such as 403), resulting in immediate failure instead of potentially very long timeouts. We understand that saying this doesn't help, but we thought it might help provide a bit more context on the issue. Unfortunately, there is no single way for us to fix this problem. We do have several ideas on how to improve this so that in the given situation, such as implementing faster connection timeouts or specifically detecting timeouts and eagerly skipping subsequent requests to the same server, such that the Package Manager does not end up blocking the Editor for so long. We plan on implementing some of these improvements gradually, but we are not being able to provide any ETA on these improvements, for which we apologize.

This report highlights the usability issues in the case of a service being unreachable that does not fail with an error immediately but instead times out eventually (depending on the default OS timeout). This situation can lead to essentially locking up the Editor as soon as there is a non-trivial number of packages in the project. Some VPN-protected servers would treat requests from clients not connected to the VPN if it didn't exist, simply not responding to the connection request. Other VPN configurations might instead immediately refuse the connection or respond with an HTTP error response (such as 403), resulting in immediate failure instead of potentially very long timeouts. We understand that saying this doesn't help, but we thought it might help provide a bit more context on the issue. Unfortunately, there is no single way for us to fix this problem. We do have several ideas on how to improve this so that in the given situation, such as implementing faster connection timeouts or specifically detecting timeouts and eagerly skipping subsequent requests to the same server, such that the Package Manager does not end up blocking the Editor for so long. We plan on implementing some of these improvements gradually, but we are not being able to provide any ETA on these improvements, for which we apologize.