0:00

Why is Python so popular? There's plenty of

0:02

room for debate on this, but one solid

0:04

reason is it's easy to adopt, easy to

0:06

use, and caters to people who are not

0:08

quite developers, data scientists, but need

0:10

to get some computing done. Do

0:13

you know where the largest untapped set of

0:15

those people hang out? Excel.

0:17

That's why it's super exciting that Python is

0:20

now going to be built directly into

0:22

Excel. Just go into a

0:24

cell and type equals pi, and you're

0:26

off writing full Python 3 code that

0:28

is backed by a lightweight anaconda distribution

0:30

of Python. We have

0:32

Dr. Sarah Kaiser here to give us the

0:35

rundown on Python and Excel. This

0:37

is Talk Python to Me, episode 445, recorded live and

0:41

on location at Microsoft Ignite in Seattle,

0:43

November 16th, 2023. Welcome

1:00

to Talk Python to Me, a weekly

1:03

podcast on Python. This is your host,

1:05

Michael Kennedy. Follow me on Mastodon, where

1:07

I'm at M Kennedy and follow the

1:10

podcast using at Talk Python, both on

1:12

bostodon.org. Keep up with the show

1:14

and listen to over seven years of

1:16

past episodes at talkpython.fm. We've started

1:19

streaming most of our episodes live

1:21

on YouTube. Subscribe to our YouTube

1:23

channel over at talkpython.fm slash YouTube

1:25

to get notified about upcoming shows

1:27

and be part of that episode.

1:30

This episode is sponsored by POSIT

1:32

Connect from the makers of Shiny.

1:34

Publish, share and deploy all of

1:37

your data projects that you're creating

1:39

using Python. Streamlit dash Shiny bokeh

1:41

fast API, flask, quarto, reports, dashboards

1:44

and APIs. POSIT Connect

1:46

supports all of them. Try POSIT

1:48

Connect for free by going to

1:51

talkpython.fm slash POSIT. POSIT. And

1:54

it's brought to you by the PyBytes developer

1:56

mindset program. PyBytes core mission is to help

1:58

you break the vicious cycle. of tutorial

2:01

paralysis through developing real-world applications.

2:04

The PyPipes Developer Mindset program will help you

2:06

build the confidence you need to become a

2:08

highly effective developer. The link is in your

2:10

podcast player show notes. Sarah,

2:13

welcome to Talk Python to me. Hey, how's

2:15

it going? Hey, it's going super well, super

2:18

well here at Ignite for one more

2:20

time to talk about cool stuff with

2:22

Python and things going on. This time

2:24

it's going to be Excel

2:26

and Python. It's like Reese's Pieces,

2:28

you got your chocolate and my

2:30

peanut butter, like what's going on

2:33

here? I mean, the cats and

2:35

dogs, are they living together or what's happening? The

2:37

old school and the new school. That's

2:39

right. The open source and

2:41

the not. Yeah, yeah. My background

2:44

is I've been a data scientist for a

2:46

long time. I did my PhD in quantum

2:48

computing in like experimental physics stuff. So trying,

2:50

I actually, my first programming

2:52

language was Mathematica and that was actually

2:54

what got me hooked on kind of

2:56

the notebook sort of interface. Yeah, okay.

2:58

Well, normally people say their first programming

3:01

language when that's the background is Matlab.

3:03

Yeah, yeah, yeah. But I would say

3:05

Mathematica is like the proper mathematicians

3:08

programming space because it does

3:11

symbolic math, right? Which is so

3:13

remarkable when you see it go.

3:15

Yeah, like that was basically like

3:17

I could visualize. The thing was

3:19

being able to do easy visualizations.

3:21

So actually after undergrad, I

3:23

worked at Mathematica for a while on their

3:25

visualization team. Really? Yeah.

3:28

What technologies was that in? Thankfully,

3:31

I didn't have to touch any

3:33

of the optimization stuff underneath but

3:35

designing new gauges and plotting filters

3:38

and stuff like that. I really

3:40

enjoy visualizing and understanding data that

3:42

way. And so when I got

3:44

to grad school, realized that none

3:47

of our instruments talked together, like everything

3:49

was an entirely separate archaic,

3:52

non-documented protocol. And we were having

3:54

to somehow like either

3:56

sit By hand and match up all

3:58

of these CSV dumps. from the different

4:00

tools and stuff like that. So basically I

4:02

started learning. Python to make our lab

4:04

work if it's know. Initially it was

4:07

like using a serial and see types

4:09

libraries actually interface with these A P

4:11

eyes but I realized that someone once

4:13

showed me a Jupiter notebooks and are

4:16

like this is our it's of this

4:18

is the way and with his Seven

4:20

Mathematica for by month exactly. So I

4:22

have been using Jupiter notebooks almost since

4:25

they came out with the fits and

4:27

really enjoy that both as a learning.

4:29

Tool A teaching tool and.

4:32

A and of reproducible science To or

4:34

brad says taking the World by Storm

4:36

Roy they were here myself ignite the

4:38

keynote. They were having basically one code

4:41

oriented demo in the two and a

4:43

half hour a keynote and it was.

4:45

Now we have a Jupiter notebook with

4:47

some Python code that's gonna do some

4:50

machine learning and you know that. Back

4:52

to cats and dogs. See, I. Was

4:55

so every course these a sweaty

4:57

doing I'm. A developer advocate at

4:59

Microsoft. On the a Python

5:02

team focusing on data science and machine

5:04

learning a sorts of Python tools. So

5:06

I get it. I get to do

5:08

all the fun stuff with Pandas, Numb

5:10

P, side Pie, and on all of

5:12

those things. And that's kind of why

5:15

when I heard about the Python an

5:17

Excel preview as like. The

5:20

men's I have done many of pandas

5:22

import from thinks that know the third

5:24

of things and was kind of curious

5:26

as to see is initially I was

5:28

like who I'm pretty cool with my.

5:30

Python and not sneer why I want to

5:32

go that way But I said to say

5:34

i think I I'm convinced them are you

5:37

have Yeah yeah because it's kind of had

5:39

a different way than a Jupiter notebook. It's

5:41

hopeless as a my as will talk about

5:44

some fun of of those as we go.

5:46

Who are you have vandalizing Python on the

5:48

behalf of my Sos to? So are you

5:50

talking to like companies that do.net in haste

5:53

do Python are talking to. Companies.

5:55

That love python and same like

5:57

a you should consider Azure or

5:59

are. offering? What does your world

6:01

look like? What do you do today? A

6:03

lot of it is doing testing. I work

6:05

a lot with the VS code team and

6:07

the Jupiter specifically like the Jupiter features

6:09

in VS code to make sure

6:12

all the notebooks work to make

6:14

sure you know packages with kind

6:16

of weird output visualizations you know

6:18

are they rendering properly? Generally I

6:20

think I work with a lot

6:22

of the actual Python open source maintainer

6:24

communities in the kind of scientific Python

6:27

space so like with the non focus

6:29

supported projects. And really just try to

6:31

kind of understand what are they running

6:33

into and as we're definitely

6:35

seeing both with all of the

6:37

LLM and machine learning stuff like data

6:39

is getting so big that I definitely

6:41

can't open it in one Excel sheet.

6:45

And so the data is you know

6:47

basically word to the it's all

6:49

native on the cloud and so there's I think

6:51

a lot of kind of the academic space that

6:53

I'm I come from of like

6:55

that isn't really part of the curriculum or what's

6:57

taught or you know people kind of have to

6:59

learn it ad hoc and so

7:02

basically I kind of like

7:04

to think of myself as someone gets

7:06

to go do all that learning about kind of

7:08

how does science and data interact with

7:10

the cloud in this case specifically Azure

7:12

and. Just trying

7:14

to help build out scenarios like build

7:17

sample apps things like that that can

7:19

make it so they don't have to

7:21

waste their PhD time trying to understand

7:23

cloud DevOps stuff. It

7:26

sounds a little bit like you know a lot of

7:28

people who are just brought into a lab and said

7:30

here you're going to learn Python and then you can

7:32

go do your your actual research. They

7:35

often sort of skim out on

7:37

like some of the very basic software

7:40

like what is get what is how

7:43

do I run this like yeah kind

7:45

of things right so maybe you run

7:47

a lot of workshops like have you

7:49

basically in my postdocs I got really

7:51

tired of my teams and groups that

7:53

I was working with not knowing so

7:55

I basically made a standard two day

7:57

workshop that was like. all

8:00

of the computer science skills you missed in

8:02

your physics degree. We

8:05

did have at one point data that just existed

8:07

only on a jump drive and that jump

8:09

drive walked away and that does not feel

8:11

super good from a research standpoint.

8:15

Yeah let's just say we had

8:17

a review of that and instituted

8:19

new policies. We decided

8:21

backups are good. It's a bit

8:23

of a diversion but I worked

8:25

on some math research before

8:28

I graduated and got

8:30

my bachelor's and it was just in C++

8:33

but we had these large

8:35

silicon graphics computers like big mainframe thing

8:37

and we had onyx workstations that

8:39

went into it. We came in one morning and nobody

8:43

could log in. The computers wouldn't respond or

8:45

turn on us. It's like a quarter million

8:47

dollar computer they got used and it won't

8:49

turn on. We're like oh boy. What

8:52

had happened is one of the grad students

8:55

non-developer types just doing their

8:57

research had

8:59

a problem with their program so they

9:01

started logging to see what was going

9:03

on and they left it running overnight.

9:05

It filled the disk to the very

9:07

last byte and apparently once the last

9:09

byte was gone the whole mainframe couldn't

9:11

operate and just nobody could

9:13

even figure out what was wrong with

9:15

it forever and eventually yeah so you

9:18

know practices. Yeah we also

9:20

had a five million dollar microscope

9:22

get bricked because a younger grad

9:24

student didn't understand that Windows XP

9:27

service pack zero should never be

9:29

plugged into the internet. So

9:33

no firewall none of that. They were

9:35

just like well it has an ethernet

9:38

port and the company said they can

9:40

you know do support over TeamViewer or

9:42

something like that so they were like

9:44

yeah. So we walked in and we had

9:46

to yank it out of the wall and we

9:49

had the police there the next day because it

9:51

had already been doing a legal financial transaction in

9:54

about 30 seconds. You're

9:57

here to arrest a microscope. Yeah I was like it's

9:59

over there. That's not a computer. It

10:01

sort of is. Another

10:04

interesting thing before we dive into our main

10:06

topic here is I'm not sure if this

10:08

is real or not. Is it

10:11

happened or did it not? This quantum work

10:13

that you're doing? Yeah. So

10:15

my PhD was on

10:17

quantum cryptography hardware. So

10:19

both there exist commercial

10:21

devices. So part of it was

10:23

finding side channels and the physical

10:25

devices that you could buy off

10:27

the shelf and then the other

10:29

part was designing satellite optics for

10:31

doing ground to satellite-based

10:34

key exchange. So yeah, that

10:36

was a lot of what I was doing in the

10:38

like prototyping things in the lab that would look like

10:41

a satellite and we'd, you know, satellite on one table

10:43

and around the other. We slowly worked

10:45

out to, we did a bunch of tests in

10:47

an airplane because turns out like putting things on

10:49

a satellite is kind of expensive and. It

10:52

just blew really high at 747. Yeah.

10:57

And weather balloons turn out to be not

10:59

super predictable in the past once you launch

11:01

them. And so if you're trying to like

11:03

aim telescopes, yeah, it kind of

11:05

ended up being messy. But it was pretty

11:07

cool. We got to go. I got to

11:09

help with the design process for these single

11:11

photon kind of collection systems that went from

11:14

literally I'm just putting a couple lenses on

11:16

the table and running python stepper

11:18

motor things in my Jupiter notebook to

11:20

something that was actually automatically tracking and

11:23

working from the ground to an

11:25

airplane. Hopefully eventually a satellite but

11:27

that's kind of out of my hands at this point. Yeah,

11:30

space takes a long time to get to. A lot

11:32

of planning, a lot of, we worked on it for

11:34

eight years and then we had our launch window missed

11:37

by a week. You know, or something like that, right?

11:39

Yeah, that's right. And as a grad student, you got

11:41

to get out of there. Yeah, you're like, I got

11:43

to get a degree. I need to eat. Yeah. So

11:46

let's get going. Cool. Well, that

11:48

sounds like a really fun project. You've learned

11:50

a lot. Whenever you're talking to

11:52

hardware, I think that's a special kind of programming.

11:55

It's not just I run the program and then

11:57

it just goes a little bit like there's stuff it touches.

12:00

Yeah, like you don't get to necessarily do your

12:02

own architectural design that you have

12:04

to kind of first observe and

12:07

usually reverse engineer what the original

12:09

device engineers came up with and

12:11

then figure out how is this,

12:13

you know, this device is streaming

12:16

base, this place, how am

12:18

I going to like, these are entirely different

12:20

like application models. You really chose the serial

12:22

part. Okay. All right, then let's

12:25

go with that. Cool. All right. Now

12:27

before we get into Excel in Python,

12:30

let's maybe talk about working

12:32

with Excel from Python,

12:34

like the reverse, right? So you talked a lot

12:36

about notebooks. And many people

12:38

go from kind of notebook into CSV into

12:41

Excel. And there's some common tools for

12:43

doing that, like even built into pandas, for

12:45

example, to be well, since of what they

12:47

can do there before they have to

12:49

necessarily decide like I need to run it

12:51

in inside of Excel. I think there's still

12:54

a special value to having structured code. Oh,

12:56

yeah, for sure. That you could

12:58

have, I mean, notebooks is like kind of

13:00

a fuzzy structured code. It's not like it's

13:02

not like a proper py file,

13:05

because you can run the stuff still out

13:07

of order. But at least within the cell

13:09

it runs within order, you know, like one

13:11

dimensional out of order, rather than like Excel,

13:13

which is two dimensional potentially, three if you're

13:15

taken to other workbooks. But anyway, what can

13:17

people do if they're not just to talk

13:19

to Excel now? Right.

13:21

I think you kind of alluded to

13:23

it. But the main way there are

13:26

a couple of packages, one is actually

13:28

built into pandas. But I think it's

13:30

a dependency on open py Excel. But

13:32

yeah, that basically, I've

13:34

been playing around with that recently to

13:36

try and see if I can get

13:38

support for the new Python based cells

13:41

in that parser. Turns out parsing XML

13:43

is not fun and not something I

13:45

like doing. But basically,

13:47

the last two days parsing Excel

13:49

are XML. Oh, my gosh, it's

13:51

not funny. And in particular, you

13:53

know, standardized document specs, because that's

13:56

basically so there are wonderful packages

13:58

out there already that do. this

14:00

for you. And so what you would

14:02

do is you basically just hand your

14:04

excel file CSV whatever to these packages

14:07

and they would return them as basically

14:09

whatever you like. Usually I go to

14:11

pandas tables because there are data frames

14:14

because why not? But yeah, so there's

14:16

basically any format that you could pretty

14:18

much want. There's probably already really nice

14:21

to import that and you can just

14:23

go right on your merry way using

14:25

your python. Yeah, you can use the

14:27

xlsxwriter if you want to like create

14:30

excel. Yeah, yeah, yeah. Right? Like yeah,

14:32

it's really cool. You can even like

14:34

make plots but you can from Python

14:37

make a plot in the excel worksheet

14:39

so that when somebody opens it they

14:41

see the plot made in excel. Kind of

14:43

crazy. This

14:45

portion of talk python is brought to

14:47

you by posit the makers of shiny

14:49

formerly our studio and especially

14:52

shiny for python. Let

14:54

me ask you a question. Are you building awesome

14:56

things? Of course you are. You're

14:58

a developer or data scientist. That's what we

15:00

do and you should check out posit connect.

15:03

Posit connect is a way for you

15:05

to publish share and deploy all the

15:07

data products that you're building using python.

15:10

People ask me the same question all the time.

15:12

Michael, I have some cool data science project or

15:14

notebook that I built. How do

15:16

I share it with my users, stakeholders,

15:19

teammates? I need to learn fast API

15:21

or flask or maybe view or react.js.

15:24

Hold on now. Those are cool technologies and I'm

15:26

sure you benefit from them but maybe stay focused

15:28

on the data project. Let posit connect

15:31

handle that side of things. With posit

15:33

connect you can rapidly and securely

15:35

deploy the things you build in

15:37

python. Streamlet, dash, shiny, bokeh, fast

15:39

API, flask, quarto, ports,

15:42

dashboards, and APIs. Posit

15:44

connect supports all of them and posit

15:46

connect comes with all the bells and

15:48

whistles to satisfy IT and other enterprise

15:51

requirements. Make deployment the easiest

15:53

step in your workflow with posit

15:55

connect. For a limited time you

15:57

can try posit connect for free

15:59

for three months by going to

16:01

to talkbython.fm slash posit. That's talkbython.fm

16:03

slash posit. The link is in

16:05

your podcast player show notes. Thank

16:08

you to the team at posit for supporting

16:10

TalkByThon. That's

16:13

been around for a long time. That's not like an

16:15

announcement. Right, people have been, there's been

16:17

different ways and they've come in and

16:20

out of support for different Python

16:22

versions and stuff. But

16:24

there are ways to kind of do that. And

16:26

I see that a lot of people are

16:28

like, I'm working in notebooks, but the final

16:30

consumer of this information is not

16:32

a data scientist or a Python person. They

16:35

live in Excel and they live in Outlook.

16:37

The only thing that can give them is

16:39

Excel spreadsheets or a PDF. Yep, that is

16:41

pretty much always the output of my

16:43

notebooks is I either export it like

16:46

the whole notebook as a PDF or

16:48

yeah, basically organizing things such that I

16:50

can use pandas or something to export

16:53

it to an Excel workbook. But it's

16:55

kind of a pain then because you

16:57

have your, there's that translation step

17:00

that you kind of have to keep going back

17:02

and forth. If you're dialoguing, somebody says, oh, but

17:04

what happens if you do this? And it's like,

17:06

okay, I gotta go rerun and remake a new PDF

17:08

for you. Could you just like do it? I just

17:10

need a quick change. Like, could you just incorporate that

17:13

like, yeah. It's not

17:15

the same change, we gotta read. Ideally, we

17:17

all have nice DevOps pipelines where you just

17:19

commit a change to source and it re-renders

17:22

all the notebooks. That's in a perfect

17:24

world. Of course, that's how it works. Obviously, that's how I always

17:26

work. It is. So

17:28

I guess, before we get into the details

17:31

of how this works, like why? Right,

17:34

because we just laid out, it is possible to like

17:36

go to notebooks, have Excel kind

17:38

of as an output and a save as

17:41

in a sense. The biggest thing for me

17:43

is like, well, it's not intended, of course,

17:45

as any sort of replacement. It really is,

17:47

as you were kind of mentioning before,

17:49

it's like a different front end sort of

17:52

thing. Like we view, there's Python scripts,

17:54

there's Jupyter notebooks, which is kind of like

17:56

an alternative front end. Before writing that Python

17:58

code, I kind of think of... using Excel

18:00

and Python like this as kind

18:02

of using a different kind of

18:04

data first. Like whenever we're thinking about

18:06

data, it's usually in some form of

18:09

grid or table or something like that.

18:11

So like being able to work with

18:14

the data kind of visually in that

18:16

sort of UI is really, really interesting

18:18

and kind of to the point that

18:20

you just made about like the end

18:23

consumer might only know how to use

18:25

Excel and doesn't know anything about Python

18:27

or doesn't know anything about running DevOps

18:29

pipeline. I didn't mean that comment either

18:31

to be disparaging because there are

18:34

people that do incredible stuff with Excel. Maybe

18:36

they shouldn't, but they can and they do.

18:39

I have in kind of delving trying to

18:41

understand what are like the weirdest things Excel

18:43

can do as I've been trying to make

18:45

demos. Like people do some weird like competitions

18:47

with Excel, like there's the painting in Excel.

18:50

Is there a flight simulator in

18:52

there? A flight simulator like you

18:54

can you know with macros, it's

18:57

very scary. So I yeah, absolutely

18:59

like I am not an Excel

19:01

pro. I am a Python developer

19:03

who uses Excel when prompted basically.

19:05

Yeah. Well, there's no

19:08

question, absolutely no debate

19:10

about it that the upper bound of what you

19:12

can do with Python is greater than the upper

19:14

bound of what you can do with Excel. But

19:16

so many people just have they have tabular data

19:18

and they need reports and understanding and change

19:20

this what happens there. Yeah. And

19:23

I think the biggest value like kind

19:25

of having played with this now for

19:27

a little over half a year is

19:29

collaboration. You don't have that kind of you don't

19:31

have to have that iteration loop between okay here

19:34

we talked about it or you know you came

19:36

and sat at my computer while I showed you

19:38

you know my Jupyter Notebook so we can make

19:40

changes like that. All of the things like having

19:43

multiple people in the Excel doc at once you know

19:45

you can just send the share link and they

19:47

can join and they can run things in the notebook

19:49

they don't have to install or set up Python

19:51

and you can just you can leave comments

19:53

like you would in Word docs so like

19:55

basically that whole kind of review

19:58

and collaboration workflow. just

20:00

turns into something that people are already pretty

20:02

familiar with with like Word docs and stuff

20:05

like that Which is a thing that Jupiter

20:07

notebooks generally don't have so I think that's

20:09

kind of the one of the most interesting

20:11

Differentiators between them. Yeah, I just

20:14

like an hour and a half

20:16

ago released the episode on Notebook

20:19

7 and Jupiter lab 4 which comes

20:21

with the real-time collaboration. Yeah, which is

20:24

really exciting So it's

20:26

on the horizon Yeah If it hasn't

20:28

been like a common thing But even

20:30

so again a lot of these people are not

20:32

gonna be in the notebooks to collaborate with you

20:35

Right, right. It's fine and you played with this

20:37

for a while how much of this is going

20:39

to be a gateway Experience to

20:41

doing more like I

20:43

couldn't remember that Excel formula But I asked

20:45

co-pilot and told me to do this Python

20:47

thing and I it totally made more sense

20:49

to me Maybe I should just

20:51

learn a little more Python and and kind of grow

20:54

beyond Excel Like what do you think of this as

20:56

a first step for many people would go? I'm

20:58

not a programmer No way if not

21:00

a geek no, but then they are

21:02

yeah They totally are like if you're playing around

21:04

with a bunch of numbers and making fancy spreadsheets

21:06

and stuff. I'm sorry You're you're nerdy

21:09

about data. And so yeah, I

21:11

think it's a really because there are lots of

21:13

really cool things You can do with Excel But

21:15

as you say like some of the plotting features

21:17

and stuff You can't do like

21:19

you can maybe do but it'd be really hard

21:21

and some it might just be straight up Impossible

21:24

to make you know, like some of the map

21:26

plot lib multi-paneled Aligned plots

21:28

with like interesting statistical plotting

21:31

things that aren't just your standard bar

21:33

and chart sorts of stuff So like if

21:35

you're trying to do some of those statistical

21:37

modeling things or the other big thing is

21:39

like the Python and Excel One of

21:41

the it has Pytorch and stuff like that

21:44

So if you want to do just kind of like

21:46

entry-level machine learning sorts of stuff and you don't want

21:48

to have to like Set up your

21:50

big data source and SQL servers, you know,

21:52

whatever what have you to work with it?

21:54

If you just got your spreadsheet of like

21:57

your home sensor data or something like that

21:59

you can actually just have it make a

22:01

regression model and you don't have to like spin

22:03

up a whole thing you know in environments and

22:05

stuff like that. When you predict the battery in

22:07

your gate will be dead next week. I'm going

22:10

to need to take an action on this. Which

22:12

would be great when then I dismiss the notification

22:14

and still don't remember to do it. Why

22:17

can't we leave the house? Well remember that notification said you're not going

22:19

to be able to open the gate in a week? That

22:21

was a week ago. Mm-hmm sounds familiar. I know

22:24

we were talking earlier that you have some cool

22:26

home assistant automation and stuff. Yeah, yeah, yeah. Awesome.

22:30

So let's just jump

22:32

into it. I mean let's talk about what is Python

22:34

in Excel? Like how do people use it? What

22:36

can it do? What can it do? In

22:39

short, basically having the Python in Excel feature

22:41

means that you can when you're in a

22:43

cell if you type the equal sign which

22:45

is usually how you start a function in Excel.

22:47

Right, you might type equal some parenthesis and then

22:49

off you go something like that. Exactly. You

22:52

drag it around. So you type equals and

22:54

then you type py open parenthesis and that

22:57

basically turns the cell now into a

22:59

Python cell. So it'll have a

23:01

little visual change. It'll turn green

23:03

on the left and there's also

23:05

a keyboard shortcut to do this. It should put the

23:07

Python logo. It should be like a blue, yellow, or

23:09

a cloud. Yeah, green

23:12

is kind of the cell color. I don't know. It

23:15

is green. Anyway, we can file feedback

23:17

on every file. Community

23:19

feedback for the Python people. We need the Python

23:21

logo in there. Let's make it happen. The

23:23

idea is that basically you can change

23:25

a cell in the Excel notebook into

23:27

one that basically is kind of like

23:29

a Jupiter cell. You can type

23:31

Python code. You can import modules.

23:34

You can write functions. You can

23:36

print things like basically anything

23:39

that would be valid in a Jupiter notebook cell. Maybe

23:41

not necessarily the magic commands but

23:44

like it's fair game. Sure. You can get to

23:46

the shell as a hacker I just do. I

23:48

did actually try a lot of that. I

23:51

should say, once you've typed in your code

23:53

there and you run the cell with control

23:56

enter, it's not shift enter because that's a

23:58

very established keyboard shortcut in Excel. Excel

24:01

which moves you around in the grid. So,

24:03

pro tip, remember it's control, enter. But

24:06

what happens then is basically Excel

24:08

connects to a container instance in

24:11

Azure which has a

24:13

standardized Anaconda provided environment. So

24:15

it's an Anaconda environment that

24:18

has a fixed

24:20

declarative set of packages which is

24:22

basically kind of what 80 to

24:24

90% of projects use. We

24:27

can talk about like which ones that are, which

24:29

ones are in that environment in a second. But

24:32

your code then runs in this, you know,

24:34

it referred to as like an iron box

24:37

in this container instance. So it does not,

24:39

there's- I don't know who you are. It

24:41

doesn't know you, well, I mean,

24:43

you are, and it's basically running a

24:46

Jupyter kernel in that box. So like

24:48

the kind of communication, if

24:51

you have multiple cells in your sheet that have

24:53

Python code in it, every time you make a

24:55

change, you know, just like an Excel would normally

24:57

kind of like recalculate everything in the sheet, it

25:00

will basically ping that iron box as

25:02

if it was kind of like a

25:04

Jupyter notebook. That's the thing that surprised

25:07

me most is that it doesn't just

25:09

run locally because you could embed the

25:11

Python runtime into Excel without any trouble.

25:13

Yes. And- In terms of making

25:16

the good path work. Yes. But

25:19

I think part of the, you know, part of the

25:21

design philosophy with this as a feature to Excel

25:24

is this is an enterprise

25:26

product that enterprises want to use and

25:28

IT admins want to feel good about,

25:31

you know, what security permissions, things, you know,

25:33

where data can come from, like

25:36

all the permissioning, you

25:38

don't want the- compliance good stuff. to go, it does

25:40

what? Yeah, no. No, we're not doing

25:42

that. And we just recovered from the ransomware thing. So

25:44

we're definitely not, we're done. So

25:46

what are the restrictions? It has no network

25:49

access. Yeah, so that iron box does not

25:51

talk to, the only thing it can talk

25:53

to is your, you know, instance of Excel.

25:55

It can receive commands, but it doesn't. It

25:57

does not touch the internet otherwise. Interesting. So,

26:00

that automatically changes maybe what some people

26:02

think like for example, Pandas

26:05

read CSV, give it a URL and

26:07

say I want the third table. Like

26:09

it's not gonna do those types of

26:11

things or you can't do request.get against

26:13

some external data. Mm-hmm. Which

26:15

initially I was like really sad about. I

26:18

understand the security things but was a bit

26:20

sad. But then some of my Excel friends

26:22

showed me because this is also kind of

26:25

a challenge in Excel too like you know

26:27

how do you get data, people have things

26:29

in SQL servers, people are databases more

26:31

generally. Every minute I want to have

26:33

it refresh. Yeah, like that kind of

26:35

it's not concurrency but like keeping everything

26:37

up to date is a really big

26:40

headache and the thing is Excel is

26:42

already very good at addressing this. So

26:44

there's something called Power Query which is

26:46

another feature of Excel and that's basically

26:48

so the scenarios like you described like

26:50

if there's a CSV somewhere in a

26:52

GitHub repo that I want to pull

26:54

in, you can use the

26:56

Power Query feature can go fetch that

26:58

and bring that into your Excel file

27:00

that then you can use with Python and

27:03

Excel. So it kind of separates that task

27:05

of getting the data to work

27:07

with kind of prize that out of

27:09

the Python piece but puts it in

27:11

much safer hands. Sure. Secure

27:14

hands I guess if that makes sense.

27:16

Yeah, restricted hands. Okay, it

27:18

doesn't have access to your account, it doesn't

27:20

have a token to your account like your

27:22

Office 365 or whatever. Yeah, it's pretty locked

27:25

down. You can't pip install things, it comes

27:27

free, free can. Yeah. It

27:29

does what it says on the tin and like okay, these are the things

27:31

I can do. Yeah, and

27:33

that again sounds kind of sad

27:35

but honestly like I've been trying to make

27:37

a bunch of samples and stuff and I

27:40

haven't run into too

27:42

many like it's got

27:44

Numpy, Panda, Sci-Fi, Seaborn,

27:46

Astro-Pi. Yeah, Astro-Pi

27:48

is pretty awesome. It's got Matplotlib.

27:51

Yeah. PiTorch and he

27:53

said pi wavelets, how's that for a

27:55

throwback to my mathematical days? That's like

27:57

fast 4a transform, don't work more complicated.

28:00

Scikit-learn, SciPy, SimPy. There's

28:03

your Matlab or your Mathematica right

28:05

there. Yeah, and others, I'll

28:08

link to the options. This

28:12

portion of TalkPython to me is brought to you

28:14

by the PyBites Python Developer

28:16

Mindset Program. It's

28:19

run by my two friends and frequent

28:21

guests, Bob Deldebose and Julian Sequira. And

28:23

instead of me telling you about it, let's hear

28:25

them describe their program. 2024

28:29

is unfolding and with it comes a

28:31

world of opportunities in the Python landscape.

28:34

Opportunities for your Python career and journey

28:36

can come from anywhere and at any

28:38

time. Are you ready to seize them when

28:40

they do? Don't let this

28:42

be the year when opportunities pass you by.

28:45

Our Python coaching is designed to

28:47

prepare you for the unexpected, to

28:49

equip you with the skills and

28:51

confidence needed to tackle any

28:53

Python challenge. Whether it's a

28:55

dream job, a complex project, or

28:58

a new venture, you need to be ready

29:00

for it. With PyBites coaching, 2024

29:02

won't just be another year. Check

29:06

out our PDM program and take the first

29:08

step towards a year filled with Python success.

29:11

Don't just dream about being a developer,

29:14

achieve it with PyBites. Apply

29:16

for the Python Developer Mindset today. It's

29:19

quick and free to apply. The

29:21

link is in your podcast player show notes. Thanks

29:24

to PyBites for sponsoring the show. It

29:28

really does cover most of, obviously

29:31

there are still plenty of cases where you

29:33

will just want to use Python and that's

29:35

okay. This is in

29:37

no way, shape, or form is

29:39

this supposed to take over your

29:41

entire Python workflow. It's really, I

29:43

still am doing my Python development,

29:45

but basically if there are people

29:48

who I know I want to

29:50

especially communicate what I'm doing to, who

29:52

that is their language, I now have a

29:54

tool that I can kind of put the

29:56

things that I would normally have in my

29:59

development environment. in a tool that

30:01

they can access and interact with too that also

30:03

doesn't feel super bad to me. Well,

30:07

what's the alternative? It's the right Excel

30:09

functions and VBA macro. Yeah, I know.

30:11

I'm not doing that. Yeah, it's

30:13

a pretty massive upgrade and I do think it's

30:15

gonna be a bit of a gateway like, you

30:17

know what? This Python stuff is not that complicated.

30:19

It's kind of cool maybe. What do you do

30:21

with those notebooks again? I think

30:23

that'll be cool. Yeah, yeah. I think

30:26

that'll happen. When I think about writing Python, I'll

30:28

use something like PyCharm, like really cool code like

30:30

analysis and then think about it like with

30:33

rough maybe running that's like an integration and

30:35

autocomplete and there might even be like an

30:37

AI magic or maybe this is happening in

30:39

VS code in a real similar way. Then

30:41

when I think about writing Excel formulas, it's

30:43

like one line across. These are not the

30:46

same. So what is it like to write

30:48

Python in Excel? Yeah,

30:50

it definitely I do generally

30:53

try to keep like when I write note

30:55

like in Jupyter notebooks I try to keep

30:57

myself small and not just have giant things

31:00

but yeah, you're totally right like even just

31:02

from the view you can't really

31:04

drag it much bigger than a couple lines.

31:06

You can expand it so it's at least

31:08

multi-line. Yeah, you can see multi-line and that's

31:10

fine and it has syntax highlighting and it does

31:13

have like tab completes. It got like the...

31:15

Does it do like the language server? Language server,

31:17

yep, yep, yep. So yeah, that's actually pretty good.

31:20

As I think was in the keynote

31:22

stuff today, there's co-pilot in Excel and

31:24

so like we'll see I think some

31:26

of these AI features also kind of

31:29

help speed up some of those things. So

31:31

like if you're like I just I want

31:33

a pandas table here and you know to

31:35

convert this part of you know the diction

31:37

the column that's all dictionaries to like their

31:39

own columns and stuff like that it'll just

31:41

write it for you and you don't have

31:43

to that editor editing experience

31:46

is definitely not the same as

31:48

you know working in my beloved vs code

31:50

but well if the keynote here at this

31:52

conference was a drinking game and AI was

31:55

a word you wouldn't have lasted 15 minutes

31:57

and it was two and a half hour.

32:00

There was so much AI this and AI that

32:02

and the AI talking to the AI and here's

32:04

and it's it's quite something But they did show

32:06

a really cool thing of like I've got some

32:08

Python code in here and I've got some Some

32:11

related data and just asking like alright help me

32:14

graph this in Python It would show you like

32:16

a seaborne graph or a matplotlib or something like

32:18

that, which is pretty excellent Honestly, yeah And then

32:20

in this case here then it would be in

32:22

a format that you could share with somebody that

32:25

they could then Excel has what

32:27

if scenario features where it makes it really

32:29

easy to be like well, you know Here's

32:31

the forecast or whatever, but what if this

32:33

change what if interest rates go up? And

32:37

you have like these little drop downs like all

32:39

of those actually work And so like I can

32:41

use cells that are like drop down data validated

32:44

drop down cells and that just feeds as

32:47

Feeds into my Python code and changes,

32:49

you know, like the theme I'm using for

32:51

my matplotlib plots or something like that Go

32:54

a little meta for a moment. Could I

32:56

use? pandas are Excel

32:59

Xls Xsl X writer whatever

33:02

the extension is to

33:04

write an Excel workbook Reggie

33:06

that itself does it has Python in it

33:08

in principle. Yes Are there just formulas or

33:11

just strings that go into formulas? Yes And

33:13

that's precisely what I've been working on recently

33:15

is to try and figure out how to

33:17

integrate into some of those The existing Python

33:20

packages because basically what you need to do

33:22

is make sure all of that data is

33:24

Inserted into the XML document model

33:27

in the right way and stuff

33:29

like it's pretty cool Like you

33:31

can take it you can take an Excel You know an

33:33

Excel notebook if you just change extension to zip then

33:36

you get to see all the XML that's inside

33:38

of it And you can just read hill you

33:40

can just see the sheets that are like here's

33:42

the the Python functions And you can just see

33:44

the strings and stuff like that So you don't

33:47

think there's anything blocking that and that's something that

33:49

I want to be able to help contribute if

33:51

I can to the community So that you could

33:53

actually programmatically write your Excel Sheet

33:56

with Python from right and so things that are

33:58

not allowed like screen scrapers or

34:00

talking to a database, you could kind of

34:02

burn that into Excel as like fix data

34:05

because then keep rebuilding that workbook or something

34:07

like that. All right integration

34:09

between Excel and Python. In Excel, I can say

34:11

I want to take the sum and I select

34:13

the big section and it's like a

34:15

1 colon a 20 and it

34:18

sums those up. But in Python, data

34:20

science land, I don't even know what

34:22

that is. So how do I make,

34:24

obviously the point having Python there is

34:26

to access chunks of data, process it,

34:28

do other things to it, turn it into graphs. What's

34:31

the interop story? Part of the Python

34:33

and Excel feature is that in that

34:35

con environment, there is a module called

34:38

Excel. The abbreviation as

34:40

it's loaded is the letter

34:42

X and then the letter L.

34:44

Basically, that allows you to access

34:47

all kinds of data and properties

34:49

of the worksheet in Python. So

34:51

that's basically whether anything from like

34:54

the cell styling to you can

34:56

literally just put Excel parentheses and

34:58

then in the string, the Excel

35:00

selection, you know, a 1 colon whatever

35:03

selections and it will just then

35:05

is already available in Python. So

35:07

there's kind of this interop package

35:09

that allows you to transition data

35:11

models between what Excel thinks

35:13

of as data and what Python thinks of

35:15

as data. Yeah and basically that comes

35:17

out as a pandas data frame, right?

35:20

Yeah in general, once you run a Python

35:23

or a Python an Excel

35:25

cell that has Python code in it, you run

35:27

it. There's a

35:29

lot of cells, sheets, code. After you

35:31

run it, you have the option of

35:34

two different output formats actually. So you

35:36

can by default, I think it outputs

35:38

as a Python object. Yeah. If it

35:41

recognizes that it's a string or

35:43

a list, it might just display that.

35:45

Or you can kind of intentionally cast

35:47

it to an Excel data type. And

35:50

again, if it can automatically understand that,

35:52

oh, this dictionary should just be in

35:55

a list or something like that. And a lot

35:57

of that functionality is Kind of provided

35:59

by... It a collaboration with anaconda. You'll

36:01

see like if you hover over if you

36:03

have it return like a a python object.

36:05

If you hover over it it'll show you

36:07

like in of the class at might have

36:09

some so you the properties and then you

36:11

could drill and and say turn know turn

36:14

the output into the slaying this property but

36:16

it shows the a little pop up with

36:18

the anaconda logue of sports stuff yeah that

36:20

that's him in or same brand in there

36:22

are no I did notice that as well

36:24

that in Jupiter I can have a cell

36:26

like say like eggs equal something and then

36:28

some other point in time who. I can

36:30

say X squared. Whatever that assumes

36:33

I'm running them. Top.

36:35

To bottom or at least those to sell.

36:37

The one that finds the axis run before

36:39

the one that uses the acts which makes

36:41

notebooks little bit non deterministic. in if you

36:43

don't just go run ourselves but it excels

36:45

exit a whole nother level ride this refers

36:48

to that was points over there and then

36:50

I goes up and then he goes back

36:52

to visit all over. how do we make

36:54

sense of like the connections? Yeah it is

36:56

something that I like always is very impressive

36:58

about Excel because. I like copy and paste.

37:00

tons of things and some all the references

37:03

stay correct. In. A logic that's not what would

37:05

happen if I respected my Python code. By.

37:07

Default the execution order those left

37:09

to right and then down. South.

37:11

That's something I got Earth I

37:13

generally. Just kind of my invention. Use

37:16

single. Columns assessed that's going down and left

37:18

to right on the number of work seat.

37:20

So to evaluate everything in the first like

37:22

the up tabs at the bottom for separate

37:24

seats dat you and go cross see as

37:26

well. Oh yeah yeah and so like honestly

37:28

one of the things that have habits I've

37:30

developed working with this is the first seat

37:32

in like the top left corner I put

37:34

any sort of like in nets are things

37:36

that blink, things I want to have defined

37:38

everywhere. just put in literally the first sell

37:41

that it could ever possibly evaluate and then

37:43

everything else from there I will be what

37:45

it's going to be. Systems for and

37:47

it's that's I make sure those things.

37:49

And actually when it spins up, there

37:51

is a pre defined kind of start

37:54

ups definition block, so it will automatically

37:56

import pandas as be deemed numb. peers

37:58

and p like so you know have to

38:00

like a lot of that kind of standard boilerplate you

38:02

don't have to but if I'm doing

38:04

a notebook where I'm doing a lot of like

38:06

second learn yeah yeah I just always want to

38:08

have that there I just got a copy pasta

38:11

for that make a incredibly thin column

38:14

that that's the view word expanded right that's

38:16

actually I've done that yeah awesome but just

38:18

like in a Jupiter notebook though that kernel

38:20

like if you can define variables like that

38:22

X and you can use that anywhere else

38:24

and then you can also just put it

38:26

in a cell and you can have Excel

38:28

refer to it and it will also know

38:30

what data is there right you might have

38:32

to do the pi or the

38:34

XL to like transform yeah otherwise then you

38:37

just use it yeah yeah awesome so you

38:39

said this runs in Azure and

38:41

it's some lockdown container but

38:44

what version of Python like

38:47

what OS is the container running what's the story

38:49

what do you know about where this actually runs

38:51

are like the to the extent that

38:53

people will care yeah yeah so

38:55

I don't know the specific version

38:58

I'm guessing it's probably 311 but the idea

39:00

is that it is a well-known

39:03

and highly descriptive

39:06

and fixed container descriptive it's it's

39:08

running I believe it's just kind of

39:10

running on a standard boon to like

39:13

Docker it's an Azure container somebody could

39:15

do a sys stop platform and yeah

39:17

get that and get it out right

39:19

interestingly yeah you can try some of

39:22

that but how they've kind of helped

39:24

lock down some of that is there's

39:26

actually hooks because I was actually trying

39:28

to do this I was trying to exfiltrate

39:30

and or escape escape the box but

39:32

they actually seen me they

39:35

have hooks in the Python interpreter that

39:37

basically can catch anything before it touches

39:39

or anything outside of Python and so

39:41

it's I mean I am not a

39:43

security expert I am not a security

39:46

expert but any like naive sort of

39:48

like ha ha I'm gonna import system

39:50

like manipulate permissions OS run open Pippen

39:52

so all of that is pretty safe

39:58

Yeah, it's gonna be. You know, a modern. There's

40:00

no Python is gonna have everything up

40:02

to date and from a supply chain

40:04

standpoint, it's all provided by in a

40:07

which they have assertions about code origin

40:09

and stuff like that. So really it

40:11

should be a reproduce the most best

40:13

definition of a reproducible environment that will

40:16

literally it. So I even in perpetuity

40:18

like of you hand the worksheet to

40:20

somebody else and you ten years later

40:23

come back to it assuming you know

40:25

thera thousand blown up presumably a run

40:27

in a container of exactly the same

40:29

description. Seat on, have to worry about

40:31

owes. Shoot I wrote this and of

40:33

her son of place on that. Now

40:35

the package was built with Python. destroyed

40:38

our allies i'm sorry ass eye for

40:40

an Opportunism weren't so many things that

40:42

soul into that seventy I know it's

40:44

wrong. Yeah like so that the statement

40:46

that I saw and sort of us

40:48

abilities as exists in work both will

40:50

still calculate against the version of environment

40:52

the worth brick was created on and

40:54

users can be prompted to upgrade if

40:56

they want like new mirror and times

40:58

in Python versions or whatever. So. Basically

41:00

there's no requirements such his to

41:02

your project that Tom opinion your

41:04

versions but at the time increase

41:07

in their kind of snapshots effectively

41:09

penned virtual moment. He and as

41:11

I mentioned there is. Kind of like

41:13

a a sixty start up script for it

41:15

that does those and have suggested imports and

41:17

you know settings for things which isn't actually

41:19

view in Excel. It shows you and it's

41:21

read only right now but we eat that

41:23

might be a thing that you might be

41:25

able to edit at the beginning so if

41:27

you know there are the same things in

41:29

that one sell that I always put at

41:31

the beginning maybe I can change that for

41:34

my notebooks sorts of things ya so it's

41:36

still and preview it's not yeah you if

41:38

I go to my Excel that I've got

41:40

and sauna my mac and just try to

41:42

type this. Is the As Nine a

41:44

like not gonna words of estimate? How do

41:46

you get it suggests? So right now you

41:48

ads in the Windows Insider program. So

41:51

you have to have a

41:53

answer sixty five certain subscription for

41:55

itself. You have to enroll

41:57

in the. insiders program which is

41:59

in If you go to your

42:01

account settings in the app right now like because

42:04

it is a gradual rollout I think the initial

42:06

public preview was only in September. Yeah, it's pretty

42:08

new. Yeah, it's it's quite new Like pretty much

42:10

every week when I start working on this, I'm

42:12

like, all right What you know does

42:15

this look the same as it did last week? So

42:18

but right now it's only available for

42:20

Windows machines It is fully intended to

42:22

be available on all platforms including the

42:25

web-based version, which will be super cool

42:27

Yeah, but yeah right now windows initially

42:30

and the feature itself is

42:32

will have some sort of cost

42:34

but that is and how much

42:37

and when is TBD

42:41

Yeah, exactly. Yeah, awesome

42:43

Well really cool to get a look at

42:45

this I think for me personally I think the

42:47

coolest thing about it is that it's gonna expose more

42:50

people to Python Give them a little bit of confidence

42:52

and then they can go only 20 libraries Oh

42:56

over on my PI that org we're just under

43:00

500,000 maybe I could expand the world of it,

43:02

right? That's awesome Yeah, and you don't have to

43:04

like all right download a version of Python

43:06

now or download a version of anaconda now

43:09

Make sure you have all the things installed make sure

43:11

your permissions things on pass like all of that

43:13

just goes away We need a virtual environment activated

43:17

It's just already handled for you and get right

43:19

to the fun stuff. Yeah, that's super cool. All

43:22

right Well, I always ask at the end of

43:24

the show for some pie pi package that you

43:26

think is cool You've come across like anything you're

43:28

like, oh that's interesting people should know about it

43:30

doesn't have to be popular When

43:34

I can't think of the name Generates

43:37

a bunch of fake data for you

43:39

to use. Okay, there's I said makaru

43:41

and makaru is the website. Yeah, Baker Baker

43:44

is good, but this it's like hypothesis. Okay, come

43:46

up with a different one though Well

43:50

one that I is near and dear to

43:52

my heart is actually called PI QIR So

43:55

one of the mentioned the quantum computing stuff

43:57

before one of the open source communities. I'm

43:59

sorry really engaged with is

44:02

actually an open source consortium

44:04

to basically create a LLVM

44:06

style machine agnostic language

44:09

agnostic representation for quantum programs. Wow.

44:11

Okay. Which is really cool and

44:13

like coming most of programming right

44:15

now has kind of was born

44:17

in Python like Python was the

44:20

host for all of this but as it's

44:22

growing up and you know kind of leaving

44:24

the nest as it were from Python basically

44:26

we need ways to interrupt like this huge

44:29

existing tool ecosystem with actual hardware

44:31

and hardware that supports things that

44:33

basically the Python like mental model

44:35

doesn't really match up like and

44:37

that's why you kind of need

44:40

domain specific languages for quantum computing

44:42

but this PyQIR package gives you

44:44

kind of an easy way to

44:46

connect and like bootstrap that interrupt

44:49

from your quantum programs that you've

44:51

written in Python to

44:53

this intermediate representation QIR quantum

44:56

intermediate representation which

44:59

is actually LLVM so you can

45:01

you don't have to figure out

45:03

how to write an export LLVM

45:05

file. Yeah. Awesome. Which is both

45:07

very fun and informative and you can do it

45:09

through Python and Rust. The

45:12

PyO3 Rust crate is my favorite. Yeah.

45:14

Yeah. So that's if you're at all

45:16

interested in quantum computing stuff and

45:19

you have a Python skill

45:21

set or interest that would definitely be a

45:23

package I'd say go check out. Excellent. All

45:25

right. Well final call action people want to

45:27

get started with stuff what do they do?

45:29

Yeah definitely. There is a great blog post

45:32

with the original announcement that's got all of

45:34

the detailed steps for how you find and

45:36

enroll in the insiders programs but other

45:38

than that there is a GitHub repo if you

45:40

do get a chance to play with this and

45:42

find all the things that you hate about it

45:44

and want to come yell at me. I

45:48

field a lot of those questions and enjoy it.

45:50

So yeah you can just come file

45:53

issues on the GitHub repo and you know like

45:55

if something isn't working the way you thought or

45:57

you find a bug that's kind of the best

45:59

way. the team is very active

46:01

here. And so that's a great

46:03

way to kind of get some good

46:06

interaction with the folks who are building this in real

46:08

time. Very cool. Well,

46:10

Sarah, thanks for being here. Yeah, absolutely. Thanks so much for

46:12

having me. Yeah, you bet. Cheers. This

46:15

has been another episode of Talk Python to Me.

46:18

Thank you to our sponsors. Be sure to check out

46:20

what they're offering. It really helps support the show. This

46:23

episode is sponsored by Posit Connect from

46:25

the makers of Shiny. Publish,

46:27

share and deploy all of your data

46:29

projects that you're creating using Python. Streamlet,

46:32

Dash, Shiny, Bokeh, FastAPI,

46:34

Flas, Quattro, Reports, Dashboards

46:36

and APIs. Posit

46:39

Connect supports all of them. Try Posit

46:41

Connect for free by going to talkpython.fm

46:43

slash posit. P-O-S-I-T.

46:47

Are you ready to level up your

46:49

Python career? And could you use a

46:51

little bit of personal and individualized guidance

46:53

to do so? Check out

46:55

the PyBytes Python Developer Mindset Program.

46:58

Want to level up your Python? We have one

47:01

of the largest catalogs of Python video courses over

47:03

at Talk Python. Our content ranges

47:05

from true beginners to deeply advanced topics

47:07

like memory and async. And best of

47:09

all, there's not a subscription in sight.

47:12

Check it out for yourself at training.talkpython.fm.

47:16

Be sure to subscribe to the show, open your favorite

47:18

podcast app, and search for Python. We should be

47:20

right at the top. You can also

47:22

find the iTunes feed at slash iTunes,

47:24

the Google Play feed at slash Play,

47:26

and the Direct RSS feed at

47:28

slash RSS on talkpython.fm. We're

47:31

live streaming most of our recordings these days.

47:33

If you want to be part of the

47:35

show and have your comments featured on the

47:37

air, be sure to subscribe to our YouTube

47:39

channel at talkpython.fm slash YouTube. This

47:42

is your host, Michael Kennedy. Thanks so much for

47:44

listening. I really appreciate it. Now get out there

47:46

and write some Python code. This

47:48

is Matlab sonata, where I'm

47:50

descriptive at home. I use the

47:52

easy method on Fuck Filming to

47:54

play a game Steam run on

47:57

the Internet mobile native practices. I

47:59

have an old computer that says,