Hey @tymsk3891,
great to hear that you are satisfied with the new Row Filter. The Row Filter in particular is one of the nodes that we deprecated, meaning that it is and will be still available in its old form. If you are using modern ui, you can just type column filter \\deprecated into the search to find the old node. But keep in mind that this node is not maintained anymore.
Greetings,
Daniel
Changing the column type via rename maybe never was such a good idea after all
I agree the new nodes UI are poorly implemented. So much wasted space and the grey text. I work with large data sets with a lot of columns and the new nodes are difficult to work with. You get the impression the devlopers just work with small data sets when designing the interface for a new node.
@DanielBog
I want to display the content of several variables. I used to use the “Text Output (legacy)” that shows a list of existing variables and provide a simple selection. Now I should use the “Text View” where I cannot find the list of variables but the help says “The contents of flow variables can be inserted by using the replacement syntax “$$[“flow-variable-name”]”.”
So, now I have to manually enter each flow variable name? And frankly, I do not know how to copy the name of the flow varible (or column header)…
Hey @spider,
Sadly yes, we already know about this limitation and already have an internal ticket to make this selection much easier (internal reference UIEXT-1678). I will put a +1 into the ticket for you.
Greetings and thanks for reporting,
Daniel
Funny, I was trying to row filter the account numbers of a table, starting with 4, so 4* in the old knode. How do you do this in the new one?
This field is numeric
I also believe a “legacy” knode with the old row filter available, or that you can find it easily, would be better …
I liked the is missing option in the new Row filter! So having both knodes available would be super…
Thanks for the suggestion. I added it to the requirements for adding pattern-matching to other types than String (AP-22915).
I second all criticisim. Using Knime for 10 years the new UI is a nuisance. After 6 months of not loocking at the new UI I did a quick check if the the situation improved, but no , its still unusable.
The first thing urgently needed is to introduce proper keybindings for table view. the help file still says Shift F6, shows the first outport, wrongly, it now opens a flow variable dialog. I hate now needing three clicks to view a port output, id needs a keybind !!!
At Knime you should do field testing with long experienced users. Dont fiddle without own work experience , which leads to destruction of functionality. In addition, if you are updating nodes do the basic UI 1o1, including check if the size of the opening windows, (some open too small by default) and critically check if pivotal UI elements have not been forgotten. It is a matter of rigorous fild testing to preserve the user experience in terms of user interface elemements.
To be precise: The two most important nodes are the column filter, the joiner. when I open the nodes on my 4k screen, they look like garbage: 20 cm height, 5 cms width, with certain text scrambeling even. Inclusion or exclusion toggle removed, is there any reason? the joiner is the same, no trustworthy default viewsize. I will not spend time in resizing ill sized windows all the time. Please fix these important elements.
Thanks for the hint, works; However, on classical view it has a confliciting key binding as Alt-ENTER opens the property page. But at least one solution, yay!.
@DanielBog The Column Renamer facelift is also very poor, see below.
It wastes such vast amounts of vertical space, and it also looks grey and fuzzy on my 4k display. See how the system dialog text at the top of the window is nice and crisp, and yet the column name boxes are HUGE and the text is fuzzy. Please fix! I just can’t fathom how this was signed off. Updates should lead to improvement, not a degraded user experience.
You’ll need to open the image below to see it at 100% to understand the scaling issue. It’s like the text has been designed to scale at HD resolution not 4k.
ps is there a fancy regex way to do this tedious renaming?
I agree. Bad layout.
For your problem use R in Knime
format_column_name ← function(column_name) {
pattern ← “OrgL(\d+)Code”
new_name ← gsub(pattern, “Level \1 Code”, column_name)
return(new_name)
}
colnames(df) ← sapply(colnames(df), format_column_name)
Or just transpose to switch column names to rows so that you can utilize more ETL functions like Regex / cell splitting and then transpose back.
Hi @J_Knime_Work , it would probably have been better to split out your “ps” question on column renaming to a new topic, as the former Column Rename node (in pre 5.x / classic UI) didn’t do regex renaming either.
For this task the Column Rename (regex) node would be your friend, and still exists in the latest KNIME version:
Replacing :
OrgL([0-9]+)(.*)
with:
Level $1 $2
would have saved you from what I agree would be “tedious renaming” 
Hi @HansDieter_Zuch , with KNIME 5.3, key bindings have been introduced for the output ports. The help file that you mention is I assume for classic UI, but in modern UI there is now a keybinding for each of the output ports. Whilst I’m not necessarily a fan of the actual keys (and the fact they’re not configurable) this is I feel now an improvement over classic UI where only the first output port had a binding.
The current set of shortcut keys for Modern UI in 5.3 is available from the following menu:
The keys for showing the specific output ports in the panel at the bottom of the application window, are:
Shift-0 (flow variables) and Shift-1 to Shift-9 for the first 9 output ports.
Viewing those in detached windows is Alt-Shift-0 (flow vars) and Alt-Shift-1 to 9 (data ports)
Thx for sharing. Great stuff. However, would be good to harmonise keybinds between the two GUI modes, as I will not use the modern UI anytime soon.
In addition, the rightclick to access the outports has no indication of keybinds and the acess to the tables is buried in a submenue. Customisation of keybindings would be great, with a keybind editor aware of the allocations.
The access to the output is THE most important “interaction” with a WF and should not be obstructed so much.
Ugly, I confess, also the very handy individual “typecast” is gone. I often struggle with variable types, and this was kind of handy.
Please let me add my two cents here: Ctrl+F shortcut.
Ctrl+F shortcut can really ease the life when one works with numerous nodes, metanodes, or components. I can easily find appropriate component in CUI, but not in MUI for such a group of components:
Thus, switching between CUI and MUI is my daily routine.
Nevertheless, happy KNIMEing!
I found this discussion by accident and let me just add that this “modern ui” is a general trend that I personally dislike with a passion.
It’s all flat, lacking contrast, everything is oversimplified. It usually looks good on a picture wireframe, with mock data, but isn’t actually better from a usability perspective.
I also feel like the people making those design choice don’t really use the tool everyday, they are “design people”.
So, anyway. It’s not just KNIME, modern UI kinda sucks in everything.
My issue was also the slow load times for these Modern Nodes on classic UI. Is has significantly slowed our team down. I have since installed the older version of Knime to move forward while we look for Knime alternatives.
