A new patch release of KNIME Analytics Platform 5.3.1 is now online and available for download .
The AP 5.3.1 changelogs list all changes in detail.
If you are deploying the KNIME Server in a production environment, you may want to wait for the first production-ready release before installing it on your production server. We always send an email announcing the first version that customers can use in production - it’s not necessarily x.y.1 - it can also be a later patch release.
Somewhat annoying this update takes an age to install, and then fails on windows due to the conda environment for one of the updated plugins running into the long filenames issue. The published “fix” at Create a New Python based KNIME Extension is a dreadful hack which by it’s own admission is liable to break an existing installation
How has KNIME gone from a robust, reliable platform to something difficult to use and unreliable to maintain?
I’m sorry to hear that you had troubles updating a Python-based extension. Even if the fix of providing a different install location for conda environments is admittedly a hack, did it help in your case?
How has KNIME gone from a robust, reliable platform to something difficult to use and unreliable to maintain?
Trust me, I am as frustrated about the long path issue as you are. And allow me to say that I am wondering how the Windows developers can get around releasing operating system after operating system without properly supporting long paths everywhere. There is a registry entry that enables long paths, but as far as I can tell it does not apply in all software (not even all Windows system tools). And the flag gets reset by some Windows updates…
If you have a better idea for us than reducing the risk of long paths issues by shortening paths, I am all ears.
I fixed it for his occasion by enabling the long paths option. How MS Win can call itself a modern OS while that is still an issue is beyond me, but that’s a different discussion for a different place In the meantime, the main fix we use is that we have removed almost all extensions that use their own conda channel in this manner from our production environment
I started using the new version of knime and I am very disappointed, the old flows started disconnecting, nodes evaporated. The worst situation is in joiner, there is no longer any information about what types of data we are combining (string, double, etc.), this is a big obstacle to work. When the files we want to combine are refreshed by someone else and some mistake may be made. I am not a fan of the new look of nodes, it is a pity that in the classic user interface the look of the nodes settings changes to “Modern UI”.
On the other hand, trying to switch to Modern UI ended very quickly for me, because it works much slower than the classic version.