If you need help - I would be able to grant you access through a remote session.
Un-installing KNIME and then Installing the new version also seems to be a workaround and is what worked for me (and others) to resolve the āBlank Screenā issue. Glad there is another topic open about this and that it is being looked into!
@leo_woerteler I did not try the new workspace approach before un-installing and re-installing, but If I re-call correctly I donāt think I was given an option to even select a workspace. After the application loaded and launched, I was immediately brought to the āBlank White Screenāā¦perhaps this is because I had selected the option to automatically open my usually workspace by default, but either way the issue is fixed for me now by un-installing and re-installing
1 Like
Hey @fe145f9fb2a1f6b,
Very good points. Let me add a few details to them:
- We will soon migrate the GroupBy and Pivot to the new UI. In this migration we will definitely also look into the RowID addition.
- For the Value Lookup we missed this opportunity, but there is already an internal ticket to correct this (internal reference AP-23876)
- Type indication is high on our list and definitely something we will add back
- Adding multi column functionality to the expression node is also on our list and we are aware that we need this functionality before it can replace the other nodes
I added the type selection to the respective ticket. - Missing Path support for expression and row filter is known, but will take a bit of time before we have it ready. We have added date&time support with this release and we will continue to add more and more over time.
- For the date&time splitter I have now created a splitter version of it. Honestly this piece we missed when migrating the rest of the date&time nodes. (internal reference: UIEXT-2842)
- What is a concurrent loop supposed to do?
- We added this feature request to the migration ticket of that node.
Typically when we migrate nodes to the new UI we look for outstanding tickets and requests and try to see if and how many of those we can implement while migrating the node. We can of course not do all the requests and there are definitely times we miss changes, but that is why this feedback is extremely valuable to us and we can improve our nodes further and further.
So thanks again and keep it coming.
Greetings,
Daniel
3 Likes
regular loop processes one record at a time.
chunk loop / group loop processes a bunch of rows at a time.
parallel chunk loop splits data into equal chunks
a concurrent loop would process always N rows in parallel and pick the next when 1 finished (similar to what you have in GET, POST, etc. nodes) and allows easier load handling or interaction with resources outside of knime that e.g. have connection limits or are subject to limitations etc.
another example: if you have 100 rows, you use a chunk loop to split it into 10x10 chunks, and then have a parallel chunk loop start creating 10 individual chunks.
the inner parallel chunk loop will always wait for all 10 parallel chunks to finish before the outer chunk can move to the next 10.
with a concurrent loop, you could have it process always 10 rows. when one row is finished, it would directly go to the next (this is useful if you e.g. wait for a external system response and have a mix of long and short running requests)
(which can be anything from http requests, commandline commands that e.g. trigger lambda functions, python scripts or else)
@fe145f9fb2a1f6b have you tried the streaming extension? there data is processed pipeline parallel, instead of node by node (which is pretty much what you describe). If you put e.g. a streamed component in a parallel chunk loop you have pipeline parallelelism as well as data parallel processing.
1 Like
yes, its possible to put the streaming component inside, but the package/row setting is an lower limit and not an upper limit.
the streaming size is the minimum of rows which should be gathered before the nodes are run.
hence, a streaming node within parallel chunks will be the same as just a parallel chunks without.
further, in case of an error, the streaming will just abort and you lose all information as streaming nodes do not carry any output.
Let me add one more point to what @fe145f9fb2a1f6b already listed: Ctrl+F search node functionality that is pending from 2017 Feature Request: Ctrl-F Find Node Dialog Box in Workflow.
Itās not a criticism, but just another thing waiting for implementation.
Nevertheless, happy KNIMEing!
3 Likes
Not entirely, as the nodes within the component will be handing finished datarows to the next node before finishing the entire node. Additionally, you can control the memory footprint by choosing the batch size, meaning you can keep more things in memory. Also, as you rightfully say, intermediate results are discarded (avoiding additional I/O in the first place). Of course you need to consider the data parallelism introduced by the parallel chunk loop in order to optimize for in-memory computation.
yes, thatās true. And you want to keep all intermediate results for debugging reasons I guess?
Iām still not 100% clear on āConcurrent loopā but I like the discussion and topic Have a great weekend!
I have version 5.4.4 and i donāt see anything when I click on āCheck for updatesā. It seems like i have to download 5.5 and remove my current installation before installing 5.5. Would that delete all my projects?
Hi @topsbog1!
No, reinstalling the application doesnāt affect the workspace(s) containing your projects. You can even have multiple KNIME installations and multiple workspaces at the same time and use them with each other however you like.
You may also want to try updating again today. Yesterday we fought with a nasty upgrade bug that broke peopleās AP installations, so we disabled the upgrade from 5.4 to 5.5 until we had a fix. We finally found one late in the evening and re-activated the upgrade path again.
Hope youāll enjoy AP 5.5!
ā Leo
2 Likes
yes the update is working today. Thanks for the details about reinstallation
1 Like
Really enjoying this update, thanks! Love the upgraded AI and new features!
2 Likes
KNIME AI Extension Guide is thoughtful and detailed guide ā¦really helpful for Naive to Nerds.
3 Likes
When will the executors beeing updated on Community Hub Teams plan?
Hi folks. If you want to get a guided tour/live demo of the new features and functionalities in KNIME Analytics Platformn 5.5, weāll be hosting a webinar to do just that on 29 July.
Hope to see some of you there!
2 Likes