Visual Studio 2026 with Mads Kristensen

Ready for the next version of Visual Studio? Carl and Richard talk to Mads Kristensen about the long-awaited version of Visual Studio. Needless to say, artificial intelligence sits front and center. Mads talks about the deep integration of AI across the development lifecycle, including code completion, debugging, even natural language querying. The conversation also digs into the role of Visual Studio as a project management tool, and its integration with cloud, GitHub, and more!

Episode Transcript

Speaker 1 (00:00):
Hey Richard, Hey Carl, what do you know?

Speaker 2 (00:03):
Well, I know that our friend Michelle Rubusta Monte is
with us to tell us about something that's going on
adjacent to DEV Intersection.

Speaker 1 (00:11):
What is it? It's cybersecurity Intersection. Let's let Michelle tell
that story.

Speaker 3 (00:16):
Hey Michelle, Hey Carl, Hey Richard, how are you.

Speaker 2 (00:21):
Tell us about cybersecurity Intersection?

Speaker 3 (00:23):
Well, so, Richard and I are partnering with the group
that does DEV Intersection and next Gen AI, and we
are putting on a new conference dedicated to one hundred
percent security focused topics. And I mean, honestly, the lineup
of speakers is incredible. We have Paula A. Jenis, who's

(00:43):
here from Poland and does keynotes all over the world
and is one of the top rated RSA speakers and
black hat speaker. We're so lucky to have her. But
she's not only keynoting, she's got a workshop teaches you
about protecting your environments against hackers and shows you about
how to you know, do attacks so that you can

(01:03):
prevent them. It's pretty cool and sessions like that as well.
But we also have speakers from Microsoft. We have we
have speakers that specialize in you know secure coding practices,
Azure security, Zuero, trust architectures on Azure UH and people
who do decision maker tracks, so things around governance policy
and you know how to how to manage and your

(01:26):
production operations keep them secure. So it's an amazing group
of speakers, really excited about it.

Speaker 2 (01:31):
And I think I can count myself among the group
of speakers there.

Speaker 3 (01:35):
Well, yes you can, that is right.

Speaker 2 (01:37):
Yeah, I'm doing a securing Blazer Server applications talk and
also I think we're doing a Security this Week live
show there somewhere that is correct.

Speaker 3 (01:48):
Yeah, we'll be recording Security this Week live. We're going
to have a great panel with some folks. The interesting
thing here is we don't really have a Microsoft and
dot net and Azure focused toecurity conference yet, so that's
the reason we're putting this on as well. You know
there are other security conferences, but they have a spread
of topics that maybe don't focus on the things you

(02:10):
do day to day. And you know this overlaps with
again our community of folks that specialize in again dot net,
Azure and yeah, they need to keep it secure too,
So with tons of talks.

Speaker 1 (02:23):
Cyber Intersection is part of a trio of conferences we're doing.
They have Intersection alongside the next Gen AI Conference all
in Orlando the week of October fifth through tenth. That's
workshops and the main conference. And you can get a
special registration code if you sign up through Cybersecurity Intersection
dot com.

Speaker 3 (02:42):
Yeah, so if you sign up at Cybersecurity Intersection dot com,
then you put in this code, so Alliance cyber three
hundred and you'll get three hundred off the entry price.
So that's a special code that only works at cybersecurity
dot com. And then you have access to all the conferences.

Speaker 2 (03:04):
Like Richard said, Wow, that's cool. Thanks Michelle. I'm looking
forward to it and I'll see you there. Hey, watch
out for falling rocks. Nice, it's dot NetRocks. Watch out

(03:27):
for falling dot NetRocks. I'm Carl Franklin.

Speaker 1 (03:31):
And I'm Richard Campbell.

Speaker 2 (03:32):
We're here for your listening pleasure again. On episode nineteen
sixty seven, this is the year we were born. Yeah,
and my wife then ye yeah, and a lot of
cool things happen. Should we do better know framework first
and then do what happened in nineteen sixty seven? Or
should we Yeah, let's do that all right, don't mess

(03:55):
with the pattern, roll the music.

Speaker 1 (04:05):
What do you got?

Speaker 2 (04:05):
I have something intriguing that I have not used, but
you know, I like to look and see what's trending
on GitHub and this one is a tool that lets
you run Windows in a Docker container.

Speaker 1 (04:17):
Oh far Oh.

Speaker 2 (04:18):
It's docer doc k u r slash Windows on GitHub.

Speaker 1 (04:24):
And let me pull the classic And why would you
do that?

Speaker 2 (04:26):
Why would you ever want to do that? Yeah? Why
would you want to do that? I'll tell you what.
The only reason I could see doing that is the
same reason i'd want to VM, which is complete and encapsulation.
But I would probably want to do it for security reasons,
you know, like I want to move all my email
reading stuff over there. But in order to do that,

(04:47):
you really have to turn off network access, well local
network access, you have to limit it.

Speaker 1 (04:53):
You can do that sort of stuff with virtual desktop too,
if you would, if it's just for that interaction. Yeah,
you know. I funny do you bring this up. Timing's
impeccable as usual, mister Franklin, because I was just having
to chat about the origins of virtualization, realizing you know,
back in the day I was using virtualization for testing installs. Right,
we'd have a baseline. This is the current configuration of

(05:14):
the machine at the company. We're gonna put the new
version on. Let's install it. Make sure it doesn't break anything.
This is back in DLL hell days, right, And when
it did you could use because you were in virtual machine.
You just rolled back to the state before tweaked the install,
figure out what you did, made mistake you and try
it again. Like it was a great testing platform. And
this is like a cool way to approach that problem.

(05:35):
Not that we have those kinds of DLL hells anymore.
Now we have new and extra special DLL health. Is
there such a thing as doctor Hell. I don't know,
Maybe it's I don't know there was doctor help, but
it's definitely new get help and yamel hell. I think
everybody on those of this show knows about new get hell. Yeah.

Speaker 2 (05:49):
Yeah, yeah, so it's interesting. I would want to, like,
I say, use it for security reasons. Vms are great,
but you know they can be slow, and if they're
not slow, they're expensive and.

Speaker 1 (06:00):
They're big, big, big bulky, Like this is just a
lean way to do this.

Speaker 2 (06:05):
Yeah, And you use RDP to connect to it, and
so it's pretty cool. But I'd just like to see
if I could lock it down and you know, use
it for Internet browsing and email and all those things.

Speaker 1 (06:15):
So interesting.

Speaker 2 (06:16):
If I do get phished, you know, I'm not going
to get ransomwared. That's that's why I would want to use.

Speaker 1 (06:23):
It, right, or you are, but it's just going to
be on a machine that you don't have anything important on,
so you don't care. Just pave it and go again.

Speaker 2 (06:28):
Yeah yeah, or you.

Speaker 1 (06:29):
Don't even have to pave it. You just rebuild the
container and to help you go.

Speaker 2 (06:34):
Right, yeah, yeah, so long as it doesn't touch your
discs and your you know, well.

Speaker 1 (06:40):
You give it a yeah, he got no rights to
anything else.

Speaker 2 (06:42):
Give it an area of the disk, and give it
limited network.

Speaker 1 (06:46):
No, it's living in that little space. You put it
on its own virtual network, so it only can communicate
the Internet, nothing else in your network. Like, yeah, we've
done all this. This is good honeypot testing and black
app testing, like this is where you do this stuff.
It's cool idea.

Speaker 2 (06:58):
So it's cool. I'm going to talk to Doane and
Patrick about that on the next security this week. But yeah,
that's what I got. So who's talking to us.

Speaker 1 (07:05):
Richard grabbed a comment off a show from fifteen sixty
going back a few years. It's from July of twenty
eighteen with one Mads Christensen. Maybe you've heard him and
can't wear his glasses very well when we're talking about
writing visual studio extensions, I'm talking about his headshot because
he's got this headshot with his glasses all askew, which
is hilarious inside joke. And admittedly this is a few
years ago, so this is a comment from Scott Hirstell

(07:27):
who said, holy smokes. After listening to this episode, I
took a dive into the visual studio marketplace to see
what kinds of extensions were available, and after hearing about
the image optimizer on the show, Thank you Mads, I
figured i'd give web Essentials a shot. One of the
most satisfying things about this tool was the ability to
save my files and have the browser automatically reload so
I can check out my changes. There are so many

(07:47):
goodies and web Essentials, thank you for your excellent work.
Oh and Fartacus is pretty good too. Everybody at the
officer you get a kick out of that one was
this from the era of the fart app was that
what Fardacus was.

Speaker 4 (07:59):
Yeah, the part is the original fart app for visual studio.

Speaker 2 (08:04):
So like you press a control key and it goes
or something or.

Speaker 4 (08:08):
Well, of course there's a hot key, but you can
also you set it up so that when the build fails.

Speaker 1 (08:13):
Ah, it's that or sad trombone, right.

Speaker 4 (08:18):
There are some forks of it that does like you know,
Homer Simpson kind of.

Speaker 1 (08:22):
Oh yeah and stuff. I remember setting up a rig
where when the build failed, it turned on a red
flashing light and everybody on the team had nerf guns
and you all shot the guy who wrote the bell.
It's nice, yeah, Scott, thank you so much. Here comment
and a copy of music Goby is on its way
to you. And if you'd like a copy of music Cobe,
I write a comment on the website at dot NetRocks

(08:43):
dot com or on Facebook, so you publish every show
there and if you comment there and I read it
on the show, we'll send you a copy.

Speaker 2 (08:48):
Music go by and we're talking about music to Code by,
which you can get at music to Code by dot net.
We have twenty two tracks all in MP three flack
or wave format, helping people write code since twenty sixteen.
I can't remember when I started it, but anyway, that's it.

(09:08):
So let's talk about nineteen sixty seven. Brother.

Speaker 1 (09:10):
All right, dude, where do you want to go?

Speaker 2 (09:11):
Well, first of all, we were born that year.

Speaker 1 (09:13):
My wife is born that year. That's ye happened.

Speaker 2 (09:16):
But one thing that's really important to me is this
is when Sergeant Peppers was released.

Speaker 1 (09:22):
Yeah, right, that's true.

Speaker 2 (09:24):
One of the most iconic rock albums of all time.

Speaker 1 (09:28):
It was also a transformative album for the Beatles too,
like it was a very different style.

Speaker 2 (09:32):
Well, they were sort of in a rut. I remember
Paul McCartney talking about this. They were in a rut
in Beatlemania, you know, and they wanted to break out
of the mop top kind of image and they were
just beside themselves. They're like, well, this is who we are.
And Paul came up with the idea and he said,
what if we just put on these personas and we

(09:53):
wrote music from the perspective of these guys rather than
the Beatles, Right, So looks end and it was really
good and it broke them through. That broke them through
their issues. So other things that happen. Of course, it
was the summer of Love Right the Human Being in
San Francisco on January fourteenth marked the beginning of the

(10:16):
Summer of Love, attracting over two twenty thousand people celebrating
piece of music. Vietnam War, Lynn and Johnson requested more
funding for the Vietnam War during his State of the
Union of Dress on January tenth. Civil Rights Thirdgod Marshall
was confirmed as the first Black Supreme Court justice on
August thirty, big significant milestone. I'll let you talk about

(10:38):
science and space, but sure, yeah. I think Jack Ruby
was shot on January third. I think you're a year
too far ahead. Oh really yeah, Oh Jack Ruby's death.
Jack Ruby, who killed Lee Harvey Oswald, died on January third. Yeah,
so he died that year. It wasn't he didn't get shot.

(10:58):
I had that backwards. What happened in space and tech?

Speaker 1 (11:01):
There's all the time. There's a lot of stuff coming
in tech at this point, like the you know, we're
only a year or two away from arpinnet becoming real,
so we're just we're just getting going there. But the
language logo is first published in nineteen sixty seven. While
if You're say gets say more, Pact and Cynthia Solomon.
They were working for a company called BBN, who was

(11:22):
also a central part of Arpanet. Sorry, the dog, after
being silent for hours, has decided to bark.

Speaker 2 (11:28):
Now, Oh that's all right. We have to be real
once in a while, don't we. Yeah, little dog bark
never heard anybody. It was logo and a particular type
of language like function, functional or procedural or.

Speaker 1 (11:40):
It was a it was a It was an interpreted language.
And this is where turtle graphics came from, like sort
of vector draw on the screen. Right, Yeah, it's turtle basic. Yeah, well,
turtle basics another thing. But this was a turtle logo.
You used turtle graphics as much graphics. We suppose to
be the term they called it logo from the Greek

(12:02):
logos for word or thought.

Speaker 2 (12:06):
Interesting. So anything in space yeah.

Speaker 1 (12:09):
Well yeah. This is the year of the most launches
in history until twenty twenty one until SpaceX blows the
record out one hundred and seventy two launches during the
Soviet Union and the United States.

Speaker 2 (12:24):
Wow.

Speaker 1 (12:24):
It's also the year of the Apollo one fire. In January,
chrisdom and White and Chaffee lose their lives in a
plugs out test on the stand and stops the Apollo program.

Speaker 4 (12:36):
Wow.

Speaker 1 (12:36):
For some time as they figure out the mistakes that
they were making. There was wiring problems. They were using
a pure auction environment, which created more risk. But it's
also the year that they fly for surveyors, which were
the robotic landers for the Moon, three of which will
landsafe on the Moon. The fourth will make a what
they call vigorous land. Also, so use one flies if

(13:00):
the first Soyuz flies with a single cosmonaut, Vladimir Komarov,
who will also lose his life after a day in
space on the way down, the parachute on his re
entry module failed. Any bomb did not survive, and the
Soviets successfully get the Venera four mission after several tries
to Venus and it lasts about ninety minutes before it melts.

Speaker 2 (13:22):
Yeah, what were they thinking, Hey, let's go to Venus.

Speaker 1 (13:26):
Well, now they were trying to You know, they taught
us a lot. The Soviets are the only nation to
ever land anything successfully on Venus. Oh, they did land,
but you know, it lasted a while. It's very tough.
Conditions are very very hot, high pressure there.

Speaker 2 (13:39):
Wow.

Speaker 1 (13:40):
And finally by the end of the year they ratify
the Outer Space Treaty, which is especially contrapuntal today now
because this is the treaty that says you can't own
the Moon and no weapons in space and don't go
setting off nukes on the moon, which is one of
the American things. Americans considered doing was to test a

(14:00):
nuclear upon on the Moon. So that doesn't happen, and
the out of space were you're supposed to stop all that.
Now there's a conversation about does this treaty still make
sense because it's from nineteen sixty seven. It's a long
time ago, and now that we're actually talking about utilizing
resources on the moon, how do we make this work?

Speaker 2 (14:14):
Very cool? Yeah, good stuff. I think this year's geekout
is going to be really good. Are the geek outs?

Speaker 1 (14:21):
Yeah, there's going to be a few geekouts, including when
after this, when this show is published, I'll have already
taken a tour of Copenhagen Atomics thorium reactors, so oh wow,
Either I'll be dead or I'll have a phenomenal interview
with their team talking about building thorium right, or both.
As it turns out, you'll publish it posthumously, right, I'll

(14:42):
actually there's no They're in Denmark and Denmark does not
allow nuclear reactors, so what a great place to build one. Yeah,
and so what I don't have a whole lot of
risk there. They're actually making it portable enough that they're
going to take it to Switzerland to operate it as
a test run. Then they're allowed to bring it back
as long as it's not operating and do some tests

(15:02):
evaluations of it. So that's what's going to be good
with that. But yeah, no exciting times. Don't worry the fall.
This winter's geek outs will be big, yeah, and there
may be a few extra ones.

Speaker 2 (15:11):
Yeah, all right, and with that, let's bring on Mads.
Mads Christiansen is a program manager on the Visual Studio
team at Microsoft, with the privilege to work with the
extension community and ecosystem. He's an avid extension writer himself,
with over one hundred published extensions to Visual Studio Marketplace.
Before joining Microsoft, he spent a decade as a web developer,

(15:31):
working at both startups and enterprise companies. His wife and
two young sons all enjoy and support his adventures in
the world of home automation. Welcome back mads.

Speaker 4 (15:41):
Hey, guys, glad to be here.

Speaker 2 (15:42):
You mad extension writer. You.

Speaker 1 (15:45):
I think we're over two hundred now, wow, crazy, got
to update your bios. You are the marketplace, sir.

Speaker 2 (15:51):
That's right.

Speaker 4 (15:53):
Yeah, I don't know. There's a lot of extensions out there.

Speaker 2 (15:56):
So far be it from me to use dot net
rocks as a place where I can get a personal
answer about some feature that I want in visual studio.
But I'm going to do it anyway.

Speaker 4 (16:05):
Yeah, go ahead.

Speaker 2 (16:06):
So I've noticed, you know, I'm a Blazer developer mostly,
and when I'm building a web application, I noticed that
every time I hit play, the position and size of
the browser that comes up is automatically set something to
where the last browser window that had focus is. And

(16:28):
it drives me absolutely nuts. And because you know, I've
got windows all over the place, and if I click
on the one to mute the TV that I'm watching,
and then I go back and press play, now it's
over the TV window, right. So I wanted to know
if there's an extension that you've created to automatically set
the position and location of a browser every time you

(16:49):
run it.

Speaker 4 (16:49):
Well, no, I have not written anything for it, but
this is I don't know how this is, like it
happens to me too when it has nothing to do
with Mitchell Studio. Right, but if you have multiple you know,
edge browsers open, you have one for your work account
and one for your private or whatever, right, and you
then click a link and outlook, Yeah, it will open
it in the last browser you had active.

Speaker 2 (17:10):
Yeah, so it's a browser thing. But surely there must
be a way to tell the browser. Yeah, open it here,
like in JavaScript or something.

Speaker 4 (17:17):
Yeah, and really you don't want to even doc that,
you know, the tap It should just be its own thing, right, Yeah? Yeah,
did you? I don't. I don't know if we I've
never seen this request before, but that doesn't mean it
doesn't exist. If it does exist, you should vote for
it or create it.

Speaker 2 (17:31):
I entered it as a suggestion in the Visual Studio.
There's a link too. Yeah I did that. I haven't
heard back yet. I just didn't know. I went looking
and I didn't see anything. But I figured, you know,
if anybody knows Mad's nos, Yeah, so.

Speaker 4 (17:43):
Make sure you put a link in so that everyone,
any listener can go and vote.

Speaker 1 (17:46):
For There you go, Yeah, leverage leverage the podcast friend
if you get it voted up to the top, you know,
then they're they're going to talk about it. Oh yeah,
that's cool, that is their priority list.

Speaker 4 (17:56):
Yeah, well, we don't know. Sometimes you see something like
this and you think, oh, yeah, this is not going
to be that hard to do, right, Yeah, and then
it turns out it's really really hard to do. Right.
So even though you are able to vote something up
to the top, that doesn't guarantee that we can do
it if it's like really expensive.

Speaker 1 (18:13):
Right, or or we don't know how well, and you
always have the niche problem, right, like a certain group
of people really want this, but it represents a tiny
amount of the customer base, Like there's always got to
be a waiting on those kinds of Well, it depends.

Speaker 2 (18:26):
I think it's one of those features that people don't
know they need it until they get frustrated enough and
say what the heck right exactly?

Speaker 4 (18:33):
And we usually say that, you know, one vote or
one bug represent a thousand people.

Speaker 1 (18:39):
Yeah, so yeah, it won't.

Speaker 4 (18:40):
It doesn't have that many people like or it's a
small niche. Let's say it's only five percent of you know,
Visual Studios user base or whatever, but that's still like
hundreds of thousands of people, right right, So it's important
that you know even the small things are big just
because we're dealing with like millions of monthly active users,
right right right.

Speaker 2 (18:59):
I heard that you guys are I don't know who
you guys is, but Microsoft engineers are required to look
at that list twice a week and actually look at
all the new suggestions.

Speaker 4 (19:11):
Yeah we so we don't really have a requirement for that,
but we are required to fix any high important bugs
like P one' or P two priority one or zero
bugs okay, and they have we have an whole SLA,
like a service license agreement there with like that we
have to fix them for the next service release that

(19:32):
we that we publish like on a weekly basis. So yeah,
we take that stuff really seriously. So you probably can't
do that without looking at your list of bugs twice
a week. But there's no requirement for looking at the list.
The requirement is to fix them. Yeah, but that's for bugs,
that's not for feature requests.

Speaker 2 (19:53):
Oh right, yeah, yeah, this is this is just the
suggestion list, I guess future requests list.

Speaker 4 (19:58):
Yeah, yeah, well we have actually do we do implement
one point two feature requests per workday.

Speaker 1 (20:07):
Wow.

Speaker 4 (20:07):
If you look at the average of the last twelve months, wow,
and that that trend is going up.

Speaker 1 (20:11):
Wow. Are you guys going faster or are they get
requests getting easier? Like, how do you get that quest? No?

Speaker 4 (20:18):
I think it's a little bit of a mind shift.
I don't know if you've noticed it over the last
two or three years. But things that before wasn't prioritized
because it was maybe had low impact, Like some bugs
come in or some feature requests come in and they
kind of have low impact. They don't you know, some
of these suggestions were not like something that would open
new scenarios. They might just be like little delighters or hey,

(20:41):
it would be nice if but they weren't like blocking
you for doing what you needed to do. And we
had a hard time prioritizing that type of feature request
in the past, whereas now it's a little different because
you know, when you're using visual studio eight ten hours
a day, you're using half your waken live in visual studio, right,
it's kind of crazy. So all these little things, because

(21:02):
you're using it so much, they become big and they're
big things. And so by kind of making that change,
of mindset a little bit to look at things differently.
We've been able to double down on a bunch of
the small issues and a bunch of the small like
paper cuts, industry standard things, muscle memory issues that didn't

(21:25):
before carry over from other apps into Visual Studio for shortcuts,
for instance, like just taking them one out of time,
Like let's just double down on that type of stuff,
and you know it. I think the shift is if
you know, this sounds super cliche, but the shift I
think has gone a little bit from focusing, you know,

(21:45):
primarily on productivity and instead of focusing more on happiness,
like developer happiness. You can't be happy developer if you're
not productive. So it includes the productivity thing, but it's more.
It's also about like are you looking forward to working
studio every morning?

Speaker 5 (22:01):
Right?

Speaker 4 (22:01):
Is there a smile on your face? Do you feel empowered? Inspired, creative?
This vis studio kind of get out of the way
when it needs to and help you when that's a requirement,
like all that type of stuff.

Speaker 2 (22:14):
I've got issues.

Speaker 4 (22:16):
We all have issues, probably.

Speaker 2 (22:18):
Probably nothing to do with what you work on, but
isn't me or has GitHub co pilot just completely taken
over like now I start typing a variable and it
gives me this long thing and there are gray letters
in there that are like and as I'm typing, like,
it's I don't even know when I'm typed, I have

(22:39):
no idea what's going on. It's gotten so crazy. Yeah,
and I know you can control that, but.

Speaker 4 (22:45):
So you can control it now. So what we're adding
is a way for that not to happen automatically, but
only at a keyboard short cut. So instead of having
it automatically show up, it can be only when you
request ith.

Speaker 2 (22:59):
Okay, I like, so.

Speaker 4 (23:00):
You you you can control more of that because some
people they really like kind of that feeling of just
tapping their way through and and other people they really
want not to have that. And so now you can
figure it. You can. You can now customize it.

Speaker 2 (23:14):
How about a foot switch?

Speaker 4 (23:16):
Yeah you can do on off. Yeah you can do that.

Speaker 2 (23:18):
I love I love foot switches. I'm a musician, right,
so I have guitar pedals and stuff, but you know,
if I had a little pedal board under my you know,
to to Oh, I want that feature. I want that feature,
these momentary on off toggles, that'd be so cool. Somebody
should do.

Speaker 4 (23:36):
That, that'd be the equivalent of a distortion distortion pedal.

Speaker 2 (23:39):
There you go, yeah, exactly. You know I'd add some
distortion to my to my API controllers.

Speaker 4 (23:44):
Yes, yeah, you know a lot of a lot of
effort goes into it. And the new version of visual
Studio here is like it's a big foundational upgrade when
it comes to to AI. So with with visual Studio,
we are in a unique position where we know a
lot about all the different aspects off your workflow. We
know about your GET issues, it's your tracker. We know

(24:06):
about profiling and debugging and unit testing, and maybe we
know about your CICD, and we know about your let's say,
ASTRA deployment environments and whatnot. And so the more we
know and the more Copiled is able to understand, the
more useful it can become. And so this is really
kind of one of those foundational updates that kind of

(24:28):
infrastructure wise sets us up for the future and for
adding features on top that's like way more that goes
way deeper and just more helpful.

Speaker 2 (24:39):
And are you guys, guidance seems to be like an issue,
But then again, I don't go looking for it. I'm
so spoiled by chatch ept searching the web for me
that I'm really disappointed when I go to Google and
I type something and I don't get like good results,
you know. So like a video that explains how to
control co pilot, I'm sure is out there. I'm just

(25:04):
conditioned to not go look for it.

Speaker 4 (25:06):
Yeah, it's out there, and a lot more will come.
But yeah, there's a bunch of great videos out there
already and articles and tutorials and whatnot.

Speaker 1 (25:15):
Oh I know, yeah that's good stuff.

Speaker 4 (25:17):
But I think but it's an interesting thing, right because
the question is always like why do you need why
do you need a description of how to do something right?
It should be autodiscoverable just by the way you design
the software.

Speaker 1 (25:30):
Well yeah, more importantly, if you know how to do it,
just do it right.

Speaker 4 (25:34):
Yeah that's another thing.

Speaker 1 (25:36):
Can we skip over? They don't explain this to me,
just do it? Phase Like, yeah, I'm tired, I got
stuff to do, Like, let's move on. Yeah.

Speaker 2 (25:42):
I mean I suppose you could use the copilot agent,
you know, or chat in visual studio to help you
figure those things out. I find that to be a
very valuable tool.

Speaker 4 (25:52):
Yeah, And you know, but sometimes the best prompt is
no prompt. So that's that's when you have a let's
say you have a button that us or it comes
up automatically, and so we have we have that in
Ryan Vischoustudo. You're going to see more of that, whereas
not the burden on prompting and knowing how to use
it is no no longer on you. Yeah, it's like
if you if you want to write a commit message,

(26:13):
there's a button there. There's a generate commit message, right,
and so you don't have to prompt, you just click
a button. That makes it easy, right. Or we just
released a new way of optimizing your code, so you
can you can just make a selection of your c
sharp or whatever or VB and then you can say, hey,
rite click and say optimize this code, and you don't
have to prompt because what are you optimizing for? Like

(26:34):
is it PERF? Is it readability? Is it quality? What
is it? You know, exception handling like or something like
what is it? And so that burden of prompting and
kind of if we can take that away and make
it so that there's no prompt at all, then you're
probably in many in many cases, you're going to be faster,
and you're going to but faster and more accurate, but
you're also going to be it's also going to be

(26:56):
discoverable that you can do this.

Speaker 1 (26:58):
Yeah, that's the challenge is how do you tell me
you could optimize that code within it out interrupting me
or annoying me in the process.

Speaker 4 (27:06):
That is the challenge. Yeah, exactly, and it becomes clippy
right otherwise.

Speaker 1 (27:11):
Yeah, that's right, that's always the dance for us. Right,
It's like I all the studio does is better than anybody.
But all those little icons and floaties and squiggles, but
all these little clues that hey I know something about
your code you may not know.

Speaker 2 (27:29):
And if you install code Rush, now you're at another
level of icon crazy, right, Yeah, talking about.

Speaker 4 (27:36):
Well, you know what they say, clippy walk so Copela
can run.

Speaker 1 (27:39):
There you go, and took a lot of flak in
the process.

Speaker 2 (27:46):
I still remember this on this quiz show wait wait,
don't tell me on NPR. You know, Adam Felber is
a comedian and he's on there, and when Microsoft announced
that they were getting rid of Clippy, right, that Clippy
was was dead Adam Felbery because hey, I see you're

(28:07):
digging a grave? Is that a business grave or a
personal grave? Can I help you dig that?

Speaker 1 (28:15):
Can I help you?

Speaker 4 (28:16):
It's funny, Richard. Maybe you remember there was a conference
where there was we had someone in the Clippy costume
walk around.

Speaker 1 (28:23):
Oh yeah, I remember that, and it was immensely popular.

Speaker 4 (28:26):
Everybody walked up in high five.

Speaker 1 (28:27):
Hugely powered. Everybody loves Clippy. But it's like, yeah, because
it's a meme.

Speaker 4 (28:31):
Yeah totally. It's like comic sense.

Speaker 1 (28:33):
Yeah, yes, except that it feels good when you punch.
It wasn't meant to be serious. I don't know why
you're using it, Like, what are you thinking?

Speaker 4 (28:40):
I use it in a PowerPoint presentation sometimes and it
throws people off.

Speaker 1 (28:43):
Well, and if we wanted to do even more O
pick Rob references. We talk about Windows Bob because at
least one of the Clippy characters was that dog from Bob,
designed by the same person, Glenda Gates. I think you
know when you can use those characters, Yeah it was yeah, Melinda, Yeah,
I don't know that she did clipp.

Speaker 4 (29:01):
You it was a wizard as well. It was one
of the characters that you could use those characters from
from back then, there was a remember for Internet Explore.

Speaker 2 (29:09):
That was the Yeah, there was a called the Microsoft Agent.

Speaker 4 (29:14):
Yeah, the Agent, Yes, and you could with JavaScript or
VBScript probably be you could, like.

Speaker 2 (29:20):
I actually used that in an early online training tool
that I wrote, which characters the Wizard. But you could
use whatever you wanted to. But the idea is you
could program him to fly around on the screen and
point to things and explain stuff. So yeah, that didn't
last time. It was fun while it lasted, but it
didn't last.

Speaker 1 (29:39):
Yeah, but you know, these all feel like attempts now
to what these ll ms can do for us.

Speaker 4 (29:46):
Yeah, it's really impressive. So we have I think there's
like a couple of categories of where this is, where
this is useful. So one is where we can learn
and we become better developers. So let me take you
a example. Is the new Profiler Agent, which is the
one feature we're building on top of this foundational infrastructure
change in Visual Studio twenty twenty six, and it's the

(30:10):
Profiler Agent is able to run the profiler in Visual Studio,
which we've had for a long time, but very few
people use it. It's kind of hard to it's an
advanced concept. It's not easy to kind of understand. But
this new agent can do it for you. And it
can write benchmarks like using benchmark dot net. Wow, it
will add benchmark or run if there's existing ones, and

(30:30):
it can then find the hot paths. Then it can
give you a laundrylease stuff. Here's what I want to
do to optimize performance. It can then simulate that with
the and then run the new and run the benchmarks
again to see what the performance benefit will be. And
then you can just tell okay, go ahead, or if
there's something that you don't understand because that's the problem.
Even if you run the profiler and it says, hey,

(30:52):
this line is very expensive or this method call is
very expensive, how do you know how to optimize it? Right?

Speaker 1 (30:58):
All right? Yeah, it's like okay, the software has criticized me.

Speaker 4 (31:02):
Now what yeah, now what yeah, exactly, it found my problems.
Now what are the solutions? Right? It didn't help me there,
but now you get that help and so you know,
it could be you know, maybe I have a dictionary
that I use a linquery on and it says, hey,
if you use a hasset, we're going to be eighty
percent faster in this particular case, and you'll be like, oh,

(31:23):
that's neat next time I run into a similar scenario,
I'm going to remember this hasset thing. And so you
learn along the way. And so I really like that
idea where the copole comes in sits beside you kind
of and and you kind of get to that destination.

Speaker 2 (31:38):
Together makes you a better developer, You become better.

Speaker 1 (31:40):
Yeah, you know this reminds me of the old optimizing
website stags. We're using the profiling tools to see that
locus of concentration, Like, hey, this is a big expensive process,
but it only gets run once an hour, so we're
not going to optimize that. But this little thing, this
settle thing, is being running four hundred times a second,
and if we can shave one hundred milliseconds out of it,

(32:01):
it's worth it.

Speaker 4 (32:02):
Now.

Speaker 2 (32:02):
Is there such a thing as an agent that understands
SEQL and SQL profiling, because I would love to turn
that loose on you know, hey, why is this query
taking so long? And what?

Speaker 4 (32:14):
Yeah?

Speaker 1 (32:14):
I think you know what index do?

Speaker 2 (32:15):
Is? What indexes do I need to add to make
it work better?

Speaker 4 (32:18):
That kind of thing that might be in for as
as a mess. Yeah, if you looked at the Latest
Days is a mess I have.

Speaker 2 (32:24):
Yeah, and I know it's based on the whole visual studio,
you know, fundamental environment, So did also have those.

Speaker 4 (32:33):
I haven't played with it for a mess twenty one
twenty two preview. You might see something more updated there
in that regard. I'm not sure. I haven't followed along
that course.

Speaker 1 (32:43):
Yeah, there's a twenty five coming that's likely out in November,
not that I know anything, but I'm speculating because that's
when they would normally ship it.

Speaker 4 (32:51):
It should be in preview by the time the airs.

Speaker 1 (32:54):
Somewhere in the quote unquote ignite time frame. And I
imagine there's going to be all kinds of.

Speaker 2 (32:58):
Imagine there's going to be more a tooling seems like,
you know that's where development is heading is in you know,
the innovation is happening in AI.

Speaker 1 (33:06):
Well, there's all the new helpers, right like we saw
this extensively with Microsoft Fabric. Not to go too far
off track here, but because Fabric touches so many different tools.
When you're doing data analytics and you've got POWERBI and
the data warehousing clients and these different storage mechanisms in
the lake and so forth, each one you don't use
enough to be really proficient with so the fact that

(33:27):
it was a copilot at each layer, it's like, what'd
you want to do? Let me help you there, Like
it's really powerful.

Speaker 2 (33:32):
It seems like a good time to take a break. Yeah,
So we'll be right back after these very important messages.
Stay tuned. Did you know you can easily migrate asp
net web apps to Windows containers on AWS? Use the
app to Container tool to containerize your iis websites and
deploy to AWS managed container services with or without Kubernetes.

(33:57):
Find out more about app to Container and aw dot Amazon,
dot Com, Slash, dot Net, slash Modernize, and we're back.
It's dot net Rocks. I'm Carl Franklin's my friend Richard Campbell. Hey,
and that's our friend Mads Christensen, the Mad indentist of

(34:18):
Visual Studio Extensions.

Speaker 1 (34:20):
And Mad's you said, Studio twenty twenty six. So is
this official? What are we expecting in the version of
Studio next year?

Speaker 4 (34:27):
No, it's out in preview.

Speaker 1 (34:28):
It is okay, so it's not next year, it's like
imminently right now. One would speculate because there's certain events
coming up towards the end of the year.

Speaker 4 (34:35):
Hey, but you know, I'm not in the business of speculation, no, so,
but you know we when did we ship visuals do
to twenty twenty.

Speaker 1 (34:43):
Two in fall of twenty one? As I recall, maybe, yeah,
so I'm just doing math. How much did the whole
AI wave derail? You guys, Like, I don't envy your
situation because you push that out in the fall of
twenty twenty one. Chat GBT is the next year, and
you know, get Hub Copilot's already out, So that's a
thing like this, It's been a it's been a while,

(35:06):
Like you're not talking four years between versions of the studio,
but they've been some nutty years, a little pandemic, a
little revolutionary computing, like wooh.

Speaker 4 (35:17):
It's been it's like a train that hits you, right,
It's yeah, it was it was like that. Yeah. We
we changed the whole ORC structure, change, all of that
changed in the past couple of years to accommodate this.
So there's no question where that this is like such
a groundbreaking change for everybody, and I think we're probably

(35:40):
on the forefront of it because we are we are
kind of the lower layers of that that enables others
to kind of use.

Speaker 1 (35:46):
A name, and you're expected to dog food the new
bits so you're seeing it before we see it.

Speaker 4 (35:52):
So all that was like really heavy. But I think
like this year when we look at like get ub
copilot adoption, like this year is where it's really exploded. Yeah,
so I think the l ms now are so good
with the you know, clots on at for another like
they're just GPT five, like they're finally at a place
Jim and I are two point five is you know?
Was also another one that was like, Okay, this is.

Speaker 1 (36:12):
Good, real good.

Speaker 2 (36:13):
So I got a story for you. I was trying
to solve a CSS problem and I knew it was
a CSS problem, but I wasn't quite sure how to
solve it because you know, I'm a C sharp programmer,
not a CSS guy. And I asked Chatchept. I spent
an hour with chatchyp Yeah, I like that he's holding
up the mugs. CSS is awesome outside of the box.

(36:36):
I set the challenge to chat Ept. I spent an
hour with chatchept, and it could not figure out, for
the life of itself, how to fix this problem. So
then I went to get hub Copilot. I just created
a new private repo in gethub with a scaled down example.

(36:57):
Get ub co Pilot scratched its head for like forty
minutes and then came back with the wrong answer. Then
I went to Gemini and I, because Gemini is like
built into Chrome, right, you can ask Gemini how do
I fix this? And it completely baffled itself. It did
not have the answer. Then I went back in the
visual studio and I went to agent mode and I

(37:20):
picked Claude Son at three point five and I said
do this, And it was literally less than three minutes
and it had figured it out. Claude's Son. It is
the bomb for CSS anyway.

Speaker 4 (37:32):
It's really good.

Speaker 1 (37:33):
Yeah, and you're sort of speaking to that sort of
new reality that just harnished the agents you got. You
want a client that has access to all of them,
and you can try your different problems on different right
on different models.

Speaker 4 (37:46):
So that's that's really nice. You can now you can
bring your own keys what we call it, right, You
can bring your own API key and plug in you know,
any other clod model or whatever. So but it's it,
you know, it really depends a lot of a things.
One is the context you give it like what does
it understand about your project? And the other thing is
the prompt, like how are you prompting it? And so

(38:08):
that's the challenge because you know at what point I'll
be becoming prompting engineers. I always think that sounds a
little bit too big for what it is, Like it's
a different way of asking a question. You know, does
that make you a prompt engineer? But but you know
there's some truth to that.

Speaker 1 (38:23):
I guess, you know, yeah, because I'm all, you know,
I do a favorite of writing, so it must be
a sentence engineer.

Speaker 2 (38:27):
Also, yeah, exactly paragraph engineer.

Speaker 1 (38:32):
Don't get crazy now trying to get there.

Speaker 4 (38:34):
So but what I what I do like is you know,
I mentioned the thing where it kind of makes you better,
you become a better developer. But the other aspect is
that it will it will allow you to do things
that you couldn't do before. So when I mentioned this
stuff about the profiler right before, like I'm not a
very low level developer, like I don't I don't go

(38:55):
into unsafety sharp for instance, right, Like I've never used
a volatile keyword. I don't know what it does, and
I'm kind of afraid to know.

Speaker 1 (39:01):
But the name tells you everything you want to know.

Speaker 2 (39:03):
That's right, stay away, Yeah, danger here across.

Speaker 4 (39:11):
So that door was closed to me, right, And but
with this this new profile agent, for instance, it can
come in and it can teach me stuff. I can
see how to do this, it can it can kick
in the door that was otherwise closed to Yeah, and
so it makes the impossible possible. And I really like
that because it's like that. And then it doesn't go there,
It doesn't open the door and walk in on its own.

(39:33):
It takes me with it, right, we walk in together,
and that I think that's the magic part. I think
a lot of people miss that. They think, oh, you know, Copila,
that's going to take over my job or whatever, but
it is. It is really a cold pilot. I think
that's important.

Speaker 2 (39:46):
Yeah, I'm looking forward to the day and I don't
think it we're here yet. But it doesn't do really
well with graphics and you know, animation and all that
stuff yet that's not really in the in the models
that I know anyway. But I'm looking forward to the
day where I can think about something that I never
thought of before doing because I know it's completely outside

(40:09):
my wheelhouse. And then and then engaging some AI to
help me figure that stuff out. But otherwise I would
I would never have attempted it. But and I'm not
talking about for professional reasons, like I would never say
yes if a customer asked me to do something like that,
but you know, for for messing around and you know,
writing some fun games or something, I would totally be

(40:31):
into it.

Speaker 4 (40:31):
Did you see the Scott Hanselman and Mark Prosenovitch they
did this shader using copony. They had no idea how
to write a shrader, and they did it. And they
could use it as a background kind of animation movie
thing that just ran in the background up the windows terminal. Yeah,
and they've never written the shad before. And so again
that was a door that was to go. They could
have learned they had spent their time doing that.

Speaker 2 (40:52):
But good stuff.

Speaker 4 (40:53):
Here we are well.

Speaker 1 (40:54):
And that's always the thing, Like you have skills, you
know you could do this, but your list of needs,
the things, you know, priority list is so long.

Speaker 4 (41:03):
That's exactly it. You know, I got kids and family
and stuff. I gotta leave work, and then I gotta
leave work. I can't just you know, keep going. And
that's the most most people are in a situation like that,
your time is limited, and so having that kind of
AI sitting there and being helpful to I think it's
really good. Another one that's really helpful is it can
increase your velocity. So I have a three hundred and

(41:25):
something repos on GIDDA right, a lot of extensions and
NU could package libraries and whatnot, and I get a
bunch of feature requests and bug reporting, and I can't
really as a single person, I can't do all of
these things. But I can have the coding D and
I can assign a task, say hey, can you take
a look at this bug? Can you fix that? Or
take a look at this feature request? And the more
you use it, the more you kind of understand what

(41:46):
it can do. And then you know, oh, this is
a great candidate for the copilot.

Speaker 1 (41:51):
Sure to do for me. If you want one hundred
percent unit test coverage, you know, you could beat your
interns as much as you want to try and get there,
but boy, the soft or knocks it out.

Speaker 2 (42:01):
Yeah that's right.

Speaker 4 (42:02):
Yeah, yeah, So I feel like you can your A.
Your velocity can go up if you if you know
how to use it in the right way. But that
now you have, but then you learn that too write.
So there's some new stuff to learn. And I think
that's important that we do it and whatever time we've got,
but don't dismiss it up front.

Speaker 2 (42:20):
One hundred percent coverage. I would set that on to
go on a Friday night and then go away for
the weekend. For some of these projects that I have,
there's so big. Well.

Speaker 1 (42:30):
One of the patterns I've gotten into now when I'm
asking the tools to write code is that they sort
of do a tdd thing. I want you to write
the test as well. And one of the effects I
found with that was that often if you made too
big of a request, it wouldn't finish it. Yes, and
so including the tests and keep iterating this until all
the tests passed. Actually create a pattern to finish the

(42:55):
code request. Right.

Speaker 2 (42:56):
I don't like, you know, dot dot dot, you know,
finish the implementation here. No, you finish it. That's what
I'm asking you.

Speaker 1 (43:06):
That's what you do, you piece of software.

Speaker 2 (43:08):
You what I do?

Speaker 1 (43:10):
Is that what you do?

Speaker 2 (43:12):
Yeah?

Speaker 1 (43:13):
Nice? But I feel like more and more we're the
project managers, right Like you're managing you only write certain
bits of code that you're like, well, the tool's going
to have at the time with this, but most of
the other code you're you're just doling it out and
checking it in. I feel like software developers are uniquely
qualified in the scenario because the tools like GitHub and

(43:33):
things like, we're used to taking contributions from unknown sources
in some respects, and what's more unknown than an LLM, right, Like,
just makes sense that okay, and it comes evaluated, you know,
run it through the process integrated.

Speaker 2 (43:46):
And let's get meta here, Like you ask the copilot
to write you something and it does a pull request
and you check out that pull request. Now you should
have an agent test that pull request. Sure, why not
have another third part? Right? And so thinking about this
in terms of, oh, it's not just this one agent
that I'm interacting with, but multiple agents that do different

(44:08):
things better than the other kind. That's this is the future,
I think.

Speaker 4 (44:12):
Yeah, we're already seeing with the mcps, for instance, where
they know now you have the playwright and it can
go ahead and unbelievable too, launch your new pro request
and the browser and test that it works. And I
mean it's pretty phenomenal.

Speaker 1 (44:27):
The thing that got me with the playwright MCP was
the ownership of the test problem on my part. Now
is a set of prompts about what I need to
valuate it on this website, And so as playwright changes
on you and the site changes on you, you don't
have to rewrite the test. They're regenerated by the prompt. Yeah.

Speaker 4 (44:44):
Yeah, yeah, that's nice.

Speaker 1 (44:45):
Yeah yeah, it's powerful, and it's just a funny way
to think about software. And I don't know if necessarily
like it's fun in the sense that I'm getting a
lot done, but it's not fun in the sense of
like that old school flowing with code thing that I
used to do once in a while. It's not the
same when you're shepherding all these tools, like you're you're

(45:06):
kind of running an interrupt driven method of coding where
various bits are coming back from different agents and you're
trying to pull them together. It's like, where are we
in the overall problem space here?

Speaker 4 (45:17):
You're definitely more of a conductor of an orchestra now
when you take really take full advantage of all this stuff.
But I don't I don't see that as Hey, the
role of the software engineer is changing. It's just that
we do we work in a slightly different way. And
so we also now orchestrate stuff, but hopefully we orchestrate
the stuff that's kind of boring or the stuff that tedious,

(45:40):
or stuff that doesn't add unique where we you know
that that we want to add unique value that only
we can do, right, stuff that's like you know, makes
us feel great and love our work, and where we
have unique insights and we can be creative and come
up with these fantastic things and then have kind of
maybe the boring parts and stuff that would be great

(46:00):
if someone else would take care of that, but we
do the conducting of that.

Speaker 1 (46:04):
That was back to the old adage, right, Like your
job as the software engineer wasn't to write code, it
was to produce solutions. Yep, right, I remember. I mean
I'm old enough down to remember switching to Visual Basic.
His guy brought real tired of battling MFC, like you
were mostly fighting with Windows, not providing solutions, and VB
took that off the shelf, and people like we're not

(46:26):
really programming anymore. It's like pretty sure, I am pretty sure.
I'm going to be crying all the way to the
bank here.

Speaker 4 (46:32):
Yeah, I forget who said it? But it was was
it Repert Scoble or like Joe's bowlski or something that. Hey,
you know, as a software engineer, your goal is to
solve problems. Yeah, the side effect is that you write
code or something like that. The by product is code.
But we're here to solve problems. And so I think
that's true in this new world as well. And I
think we're still going to write as much code. We

(46:53):
might even write more because some of this other stuff
can be sort of automated away a little bit. And
so I think we're looking at a very very bright
future for software engineering. I think it's going to be
more fun now than they has been in the past.

Speaker 1 (47:06):
The productivity explosion is astonishing. Folks have seen who figure
this stuff out. I mean, I for a while there,
the good Hub Copilot days, I was seeing folks saying, hey,
my coders are pushing in more code. It's twenty thirty
forty percent more productive. There's more reworks, like they're backing
out stuff more often. But generally if it gets pushed
and it sticks, it's pretty good in its days, Like

(47:28):
code quality up overall. Yeah, Yeah, Now I watch guys
that are knocking out two three weeks worth of work
in a typical pattern in a day. Sure, Like it's
just astonishing when they get it right.

Speaker 4 (47:39):
Yeah, yeah, right, And so I think like, as, but
that's a craft too. How do you work with the
prompting and with the yelms.

Speaker 2 (47:47):
Yeah, So also staying current on what the different lllms
are and what their strengths and weaknesses are. I think
that's a big challenge too, because you know that experiment
that I did with the first chatchpt for an hour,
and then you know that the GitHub copilot and all
that stuff that took like a whole day just to
figure out that claudees on at three point five was

(48:07):
was the was the model that I wanted. And now
when I go back that, I'm going to try to
use that for a similar problem. But something else might
have come around that makes it even easier. Right, So
keeping up with these things is going to be a
full time endeavor.

Speaker 1 (48:24):
I remember the arguments over is garbage collecting actually a
good idea? We've been to these sort of changes in
the way we think about code things before. Yeah, and
it's just this is an issue one.

Speaker 4 (48:35):
And I also feel like, yeah, you have to keep
up today with like which models are great for your scenario.
Let's say, I don't think that's different than keeping up
with other industry kind of.

Speaker 2 (48:44):
Yeah, it just most faster, most faster than language evolution
or tool evolution.

Speaker 1 (48:51):
That may only be true right now.

Speaker 4 (48:53):
Yeah, it's a moment in time. I think it feels
like we're on our way somewhere. Yeah, and we're trying
to figure out where where does this go? Is this
is the chat the right interface going forward? Is it
a what is it? It might be, but it might
not be. And so we're on our way there wherever
there is.

Speaker 2 (49:09):
I have a suggestion to people who are hearing this podcast,
and that is listen to more dot ne Rocks episodes
because we'll keep you up to date on these things.
Because we're doing the hard work out here.

Speaker 1 (49:22):
We're definitely asking some questions anyway and trying to get
an overview. Yeah, that's true of you know, who's succeeding,
who's struggling? Where are these things helping? Word? Don't they help?

Speaker 4 (49:31):
But we're thinking about this. You know, you might have
heard this concept of ambient AI. Yeah, so the concept
real briefly is that it's kind of sitting there in
the background only showing its phase when it's needed, and
it seamlessly offer you, you know, the help contextually where
you want it. And so it's so think of it
as something that's always there, but it's not in your way,

(49:54):
and it's not something that you need to learn. And
so we like that idea because it kind of fits
naturally into things that already exist. So if we were
to put it into Visual Studio, for instance, you know
I mentioned the you know, the generated commit message, Well,
we already have a commit message box. So adding a
button that automatically does that would be it's a place

(50:14):
where you expected, it's not in your way, you don't
have to prompt anything, right. That's a that's kind of
ambient AI thinking. Another one is renamed variable. We made
a big splash about this because you know it's the
one of the hardest problems in computer science is naming things.
And so if you want to rename a variable, there's
like a little UI and Visual studio for that. If
you could click control r R on an identifier, the

(50:36):
rename thing come up. We show you automatically the top
three things that we think based on how that variable
is being used in your code. What will be good
names for that based on cipical naming strategies, and so
that's ambient AI.

Speaker 1 (50:48):
Although we are talking about Microsoft here and you guys
have a record about names.

Speaker 2 (50:57):
All right, well let's just put.

Speaker 1 (50:58):
That right out there.

Speaker 2 (50:59):
No size, that's the perfect name for what it does.

Speaker 1 (51:03):
You know exactly what it does.

Speaker 4 (51:04):
I'm the worst of naming things. They just the name
is what they do.

Speaker 1 (51:08):
Yeah, yeah, which is actually the perfect name. Like, because
nobody's confused in.

Speaker 2 (51:12):
The in the UK, they say, it's what's on the tin, right,
what's printed on the can is what you get. Yeah,
what's on the tin.

Speaker 1 (51:20):
It's what is in the can. Like, that's as much
as we could hope for. I appreciate that.

Speaker 4 (51:26):
Uh.

Speaker 1 (51:27):
Can we talk a little bit about the sort of
project side that is also visual studio because I think,
you know, people will tell me, you, well, why would
I use visual studio code of visual studio? And it's like,
to me, studio is as much a project management tool
as it is a development environment.

Speaker 4 (51:42):
It is. Now, that's funny, that's exactly how I look
at it too, and that relationship you have between projects
and solutions. But if you're new to visual studio, you know,
a lot of people don't understand that concept, yeah, because
it's just new to them, foreign to them, and it
really allows us to do some interesting things when it

(52:02):
comes to building and to building context, I mean compiling
and building context for intellisents and so on when it's
not based on a file system, but it's based on
an intentional project structure. So I guess you could replicate
that in the filesystem too, but that's typically not what
you see, and so that's you know, we've it's always
been a little bit problematic too, because loading from a

(52:24):
file system is fast, but loading from a project hierarchy
that you have to read the entire hierarchy first before
you can figure out the dependencies between them and how
to then render them in the solution explore right in
the tree view. And so there's always it's always been
more expensive from that perspective working with the projects and solutions.
So this time around we've we've finally i'd say, been

(52:47):
able to go really really deep on some performance stuff
to make that, you know, faster than ever. So you'll
notice the first thing you'll notice in the new version
is how fast things show up, how fast project solutions
load all that, it's just it's visible immediately.

Speaker 2 (53:03):
I was under the impression that the reason why visual
studio proper is slower than say visual studio code is
because there's a lot of calm involved in stuff, and
so that naturally kind of slows down the whole process
and kind of bloats the memory. But I don't know

(53:24):
if that's really true. What do you think about that?

Speaker 4 (53:26):
No, that that had that has some truth to it,
for sure. And but also like visual studios over twenty
eight years old, and so there's there's some things that
have just evolved over time, like some debt that has
been building up, right, and we've finally been able to
look at that, but some very key aspects of it.
But one thing that was was a thing was blocking

(53:47):
the UI threat. So when you would click a button
to do it something, sometimes whatever would happen when you
click that button would do that on the UI threat.
And what that means is that visual studio freezese. Nothing
could update on the score, right, and so we would
and so we get what we call it's called a hang,
but you might call it a freeze, and Zoo would

(54:07):
be unresponsive for like a half a second.

Speaker 2 (54:09):
Once the programmers know all about that though. I mean,
if you've done any kind of multi threaded or a programming,
you know what that is.

Speaker 4 (54:16):
That's top programmers.

Speaker 2 (54:17):
Richard had this comment about outlook sixty five threads and
not one of them for me, that's my thread.

Speaker 4 (54:29):
But when you start but when you start doing that,
and that's part of that profile the thing too, right
or or debugger where like something like an AI can
help you. Is it's really hard to d book like
acinc call stacks and you know, multi threaded applications or whatnot.
But this time around, we've really really done more than
ever in terms of moving things out of the UI

(54:51):
threat and that we run dot net you know core
processes under visual Studio that you know runs very qui
and very fast, and it does a lot of that
work in the background now. So you'll see visual Studio
becoming a lot snappier. So opening startup, opening a solution,
time you hit five, like compilation and time for you

(55:13):
to hit When you hit a five too, you hit
your break point. Is that was so much faster than
ever was and it does all that while the UI
is still respond here.

Speaker 2 (55:20):
If you guys at Microsoft, the people that are working
on Visual Studio in general are using AI to help
them find bottlenecks and find places that need to be
fixed or updated.

Speaker 4 (55:33):
Yeah, we're using it internally. You have to you have
to think about the So the Profiler Agent is really new,
So we haven't used it like extensively, like for a
long period of time. It's very new, and I tell
you it was. Some of the first things we did
was we told all the teams, hey use it on
your own code base inside Visual Studio, and so we
see optimizations come in all the time. But what Nick did,

(55:56):
he's on the profiler team, he did something really interesting.
He started with the top hundred Nugat packages and he
would profile them to figure out where are the hotspots
or whatever and can he fixed them? So he sent
starts sending pull requests to the top Nugat packages because
they're used in thousands and if not millions of apps, right,
and that means if he can like do seventeen percent
better performance for some of these packages, like that's a

(56:19):
huge win for everybody.

Speaker 1 (56:21):
Yeah wow.

Speaker 4 (56:22):
So he's been very busy doing that and that's really
great because that trains him in using the product and
figuring out where there could be further improvements and so on.
Plus all the internal dog footing stuff off that profiler
ends up making everything better and we can fine tune
all these things and it's it's really impressive.

Speaker 2 (56:41):
Well, I guarantee Polly isn't one of those newcat packages
that needed improvement.

Speaker 4 (56:47):
It's really should I tell nick to go look at that.

Speaker 2 (56:50):
No, it's really fast, it's really good.

Speaker 4 (56:53):
Oh, it's fast enough, it doesn't need Yeah, it's.

Speaker 1 (56:55):
Like all native the as your team got involved because
they were leaning on it. Pretty Yeah.

Speaker 4 (57:00):
I think actually Nicky found that when he ran some profiling,
he figured out all the way down in dot Net
framework itself that there was something that could be optimize.
So I think he's gonna send up a code update.
I'm not sure they take pull requests in the traditional sense,
but like he's gonna see if we can get some.

Speaker 2 (57:19):
That's so cool.

Speaker 4 (57:20):
That framework, because that has to do with visual studio startup.
Visual Studio actually is a native process that boots up
the dot Net framework in its own. It's like it's
really kind of.

Speaker 2 (57:30):
And you're talking about you're talking about the Donet framework,
the Windows version of dot Net framework.

Speaker 4 (57:35):
Yeah, framework for it Yep.

Speaker 2 (57:37):
That's so cool that you guys are doing that, and
it's it's inspirational.

Speaker 1 (57:42):
Really well, it's just a reminder that Microsoft employees use
Visual Studio like the products you build the products with
the products Uber Dog Souper.

Speaker 4 (57:51):
That's my joke always like, hey, did you know that
Visual Studio twenty twenty six was build using Visual Studio six.

Speaker 1 (57:57):
That's great, it's true. So where can folks take out
preview out firs spin.

Speaker 4 (58:01):
Yeah, go to visual studio dot com, click the download
button and select the preview of twenty twenty six. That's
how you do it.

Speaker 2 (58:07):
But if you already have the Visual Studio installer, you
can just load run that and it'll tell you if
there are preview versions, right.

Speaker 4 (58:15):
You need to maybe if you have the preview already,
you've got the latest on the preview that includes the installer. Otherwise,
just go to visual studiot com click it there. If
it doesn't show up for you, and you can install
a side by side. It won't interfere with any other installs.
You got a Visual Studio twenty twenty two or whatever,
and it will it will look at twenty twenty two
if you have that installed. So when you're installed twenty

(58:36):
twenty six, it looks at twenty twenty two, and it
looks at your settings, your extensions, and the components the
workloads like if you're if you've taken like let's say
dot net, ASP, dot net and desktop, those are your
two workloads. It will copy that information. It will take
that information from twenty twenty two and pre select all that.

(58:57):
So your installation is super smooth. You get all the
components you need, settings and extensions because the extensions are
you're twenty twenty two. Extensions work in twenty twenty three.

Speaker 2 (59:07):
You go, let's hear for extensions. Yeah, going into the future.

Speaker 4 (59:14):
Very exciting stuff and that. Yeah, so it's super fast
to get from you click the install butt until you're
able to open till you're able to open your solution
and start writing code. Is probably the fastest ever.

Speaker 2 (59:25):
Fantastic nice All right, Well, geez, I think I ran
out of questions. You got any more, Richard.

Speaker 1 (59:31):
No good. I think it looks like it's gonna be
an awesome version of the studio, and it's been a
while coming, but I understand why.

Speaker 2 (59:37):
Matt, thank you so much. It's been awesome as always.

Speaker 4 (59:40):
Thanks for having me. It's my pleasure as always.

Speaker 5 (59:43):
All right, and we'll talk to you next time on
dot net brocks.

Speaker 2 (01:00:07):
Dot net Rocks is brought to you by Franklin's Net
and produced by Pop Studios, a full service audio, video
and post production facility located physically in New London, Connecticut,
and of course in the cloud online at pwop dot com.
Visit our website at d O T N E t
R O c k S dot com for RSS feeds, downloads,

(01:00:29):
mobile apps, comments, and access to the full archives going
back to show number one, recorded in September.

Speaker 1 (01:00:36):
Two thousand and two.

Speaker 2 (01:00:37):
And make sure you check out our sponsors. They keep
us in business. Now go write some code, See you
next time.

Speaker 4 (01:00:44):
You got Javans

Speaker 1 (01:00:48):
And s