Databricks Notebooks: New Cell UI Debugging & Result Table

0:00hey today I have a special guest I met few months ago back in Paris so here we are with Jason hi Jason hey everyone super glad to be here thanks for having me Yosef so can you tell us more about your role at data breaks yeah for sure I am uh the product manager for the notebook and uh

0:22so I've been working on a lot of the new notebook experiences that we've been releasing recently like um the new cell UI and the new result table as as well as the debugger amazing so please show me all the new features okay awesome yeah happy to do so let me uh let me share my screen so to be honest I was really

0:45really Amazed by all the new features that has been released um I mean the it's one of the fastest product I've ever seen like you know growing with those number of features yeah no we've been doing a ton of work here we've putting a lot of investment into the notebook to just uh make it better and make it make it awesome and so um I think a lot of the

1:08stuff I'll talk about today is kind of how we've been redesigning the UI basically from the ground up to make it uh just easier to use more uh delightful to use um so as you can see like right here we have the new cell UI and the new result table um enabled and these are two recent new features that hopefully everyone has seen and got a chance to try

1:30and there's just a lot of nice things with them so like for example um I don't know with with the old new with the old cell UI it was always really complicated to move sales around and so now with the new CI you can just like click this and distract you know and it's really easy to do this um here's a little hidden trick a lot of people don't know if you want to collapse the results you can just um double click to collapse the results which is nice we also have uh

1:53now a single click to run button so you can just say run sell um and it'll do that uh we also also have a new uh Focus mode so if you just click this mode right here you can just look at the cell that you're working on so it's nice I don't know about you but I get very distracted by everything on my page and so this is nice just to focus in on exactly what I'm what I'm working on um

2:16and if we uh if click down here we'll get an output um again of uh of what we what we were running um it's also nice you can navigate this way so if I click through here I can go to the next cell um and I can continue my notebook so if you prefer this kind of uh one solid a Time Focus mode experience we have this uh which is which is hopefully going to be um awesome for

2:39folks anything uh anything you else anything else you want to call out ysep about the new ca anything you've liked in particular I believe also the possibility to rename the celles without having the without the mark down yes am yeah yeah yeah that that's really amazing and also Al the new uh let's say

3:02the um the new AI experience where we just need to add slat do and then it will add automatically the documentation to your to your code or fix the issue so this is this is really amazing I mean just clicking near the near like in this part and that's it SL command do and you have this this command generated for you so that's really really

3:26amazing yes yeah Y and it's putting comments everything um yeah I I will say here I'll go I'll go reject that because I don't want to fix this because this is this is part of my my upcoming demo so let me reject this but um you know one other thing I think with the new sui which makes it feel awesome and brand new is um we have custom theming now so if you see like like oh this is not what I usually see the of the color scheme of datab brick so if you go to view and

3:49then you go to editor theme you know this this would be the data Brick's default we can look at this there's more stuff to look at um and I can just kind of um customize whatever I want so like I like Au light it has this nice orange feel but there's a lot of other themes like this this one's a nice one um so you can you can play around and customize your your UI as well

4:12amazing cool well then that so let me go back to my my preferred theme and then the next thing that we've talked about or that I want to talk about is the new result table so again from the ground up kind of totally change the way that results um the result table works to make it better for exploratory data analysis so as you can see here I have this um playlist which talks about the top Spotify songs for 2023 and if I

4:35click into the new result table I have kind of these Excel like ex uh selection mechanics so I can just grab multiple things you know if I want to copy um I can just you know copy that really easily across multiple things which is nice if you want to do partial selection you can just double click and so then I can grab um you know like just this this this portion of it which is great um we

4:57have some data type icons here so if I hover over this we see this one is a long whereas this is a string uh which is nice just for your your analysis um if you want to sort again really easy to do a single click sort between different things here and we also have some powerful stuff that that was not available before which is like like search for example so if I want to search um all the columns for new genes

5:20you know that's a an artist I've liked recently um I I can do that which is great and then also we have specific um filtering on columns and values so you can you can add a filter by um clicking in here and then doing this or one of my personal favorite ways is by um if you right click on a sell you can say filter by this value and so now I have all the artists with Taylor Swift um and then if

5:44I for example if I want to do this again with release year of 2019 I can filter by this value and so I can really really quickly do exploratory data analysis um and then we'll add another filter just for for good measure just to show say if I want to get this cruel summer um I just add here and I just type summer and uh we'll get the one that that contains that so really really powerful way to do a bunch of exploratory data analysis with the new result table I

6:08wasn't aware of this one well there you go yeah so you just you just toggle it right here you can just uh you can go in and out so again yeah here's here's the old table very basic functionality and if you just go here and you just say um toggle it on and we're we're looking at all the feedback making law of changes and improvements so if there's anything you would like to see added or changed uh definitely let us know in the feedback in um if you give it to us I I will read

6:31it so amazing thank you thank you thank you Jason yeah and I also I've heard that you're gonna show me the new Dean experience yes I I am yes exactly so so okay so let's get to that so I started with SQL here just to show you this um and then in datab Brick section we have the ability to go from SQL to uh a

6:54python data frame or a panis data frame and so um you know I can see the output was stored in SQL DF so I'll just run this to convert this to Python and then I want to create a visualization which shows um the average veilance of different songs in this playlist but I have an error right here so I'll just I'll run this again just to make sure everyone can see um we we have this

7:16error it's not quite doing um what I'd want to and this is actually one where you know I tried the assistant and uh it didn't quite um it gave me a solution that was right but it wasn't a solu or that got rid of an error but it wasn't a solution that gave me the output that I was looking for so this is a situation where I really want to debug in my code and see what's wrong so I'll just click here to add a breako I guess I'll do here because there's actually code there

7:39and if I just click this little drop down here with the new cell UI I can say debug sell what that will do is it will um open up a uh a uh my variable window over here and then I get a little debugging frame where I can step through um just right here in this cell so I can just continue next line okay so these looks like it's it's doing what I want

8:01um you know I have I have my data frames over here those seem like they're doing what what I want them to do uh if I go to the next line let's see the Spotify average veilance mean um 60.6 that looks like it's calculated correctly apple looks like uh it's calculated correctly at 54 and then if I go over to the next line um I can see the dieser average

8:24veillance oh okay well this is not what I wanted to uh to be seeing here I wanted to be seeing a number and it looks like I forgot to call the the mean function here so I can uh I can make that fix here I can stop debugging I I could go through it again uh but in the in the just for time sake I will uh I won't I

8:47won't put you through that and we'll just run this and uh show you that now we have H have the right answer so we have yeah step through stepbystep debugging now which is a really powerful way to uh yeah step through and debug your code so this is something we've heard a lot of people want and requests and so we're excited to be bringing it uh very very soon to everyone yeah it's amazing and and also something uh

9:10which can be part of this Dean experience if remember like just few months ago let's suppose you did Run 10 qu 10 cells and then at the cell number I 11 or 12 you faced an issue so you should have uh run each cell uh alone now you just click on this run all above it will run all the sales and it will stop on this sale where you have the

9:32issue run the debug experience fix the issue and then you can keep running all the sales above and it's really quick way to improve your productivity and fix your issues quickly and put your workloads on production yes no ex exactly yeah you you said it better than me you're yeah exactly so it's really really amazing

9:57so I think I I I really can't wait to test all those uh new features and thank you Jason for running us through those uh this new experience and uh any other comments before we finish this session no I mean I guess I just want to say thanks to all of our our users honestly uh it's it's it's thanks to

10:20your feedback that we're able to make such awesome products so um continue to give us feedback and we'll continue to read your feedback and and make adjustments so um so yeah no thanks thanks for having me and I'll stop sharing now thank you Jason it was a real pleasure and see you soon