Homebrew Turns 15

21 May 2024

Interviewed by "Homebrew Turns 15" stream

Homebrew was created on May 20, 2009. Fifteen years later it’s in use on tens of millions of devices.

Show transcript

0:00 hello and welcome to the live stream homebrew is turning 15 happy birthday homebrew
0:05 my name is john britton i'm the co-founder and ceo of forkbrew and today i'm joined by
0:10 two outstanding guests uh max howell
0:14 welcome to the show hey thanks for having us let's see if i can do this there we go
0:24 so uh hello uh let me give it a short introduction so uh mike is the uh project leader of homebrew
0:32 my co-founder and cpo at workbrew and max is the original creator of homebrew um welcome
0:40 and maybe you could each give a little introduction about yourself uh max why don't you go first
0:44 uh thanks john and thanks for having me on uh good to see you again mike nice tank top
0:54 yeah so uh i'm max how i've uh been in open source for more than 20 years i stopped counting at some
1:03 point i got involved in open source because uh i thought i wanted to be a scientist which turned
1:10 out to be incorrect as everybody else told me in my life chemistry is super boring i i thought i enjoyed
1:18 it but just turned out i like the academic side of things didn't like actually working in it so one day
1:23 in a depressed funk i went home and installed linux on my home computer and discovered the open source
1:28 communities of the world this was like 2003 2004 it's about then i met mike actually we were both
1:35 working on kde at the time uh which is this desktop environment for linux which i think still exists
1:42 i don't know i haven't used it in a while but uh i'm just got me a job in the industry lost fem in
1:47 london that's where i created homebrew um back in 2009 ish so yeah 15 15 years about 11 hours ago
1:55 almost 11 hours ago apparently my commit was four minutes past midnight
1:59 yeah well mike was saying the other day how when we uh started working on homebrew like debian was
2:06 15 years old and uh that felt ancient and here we are 15 years so yeah thanks for having me on looking
2:13 forward to this uh this whole chat yeah thanks max yeah mike why don't you go ahead and introduce
2:18 yourself and kind of your relation to homebrew yeah of course um yeah i feel honored to lead max not
2:25 just for his contributions to homebrew but his contributions to excellent voices whenever i i forget
2:30 because we normally talk of our text message nowadays whenever i hear max on a podcast or whatever i just
2:35 think he is most delightfully deep voice in the world so i'll do my best to have a suitable timber uh yeah
2:44 so i'm mike uh i've worked on open source not as long as max but probably you know like sub 20
2:51 slightly over 15 years um i got into software i guess the more traditional route just going and doing a cs
3:01 degree and got uh fell into linux distros and all that nonsense when i was at university and ended up
3:09 with one commit in the licks kernel which is always a nice kind of little humble brag when people accuse
3:16 me of being one of the mac people who doesn't know anything about linux um so yeah i guess i uh also
3:22 ended up in london at the same sort of time as max i worked for a startup called mendeley uh down there at the
3:28 time and i was kind of trying to wrestle with mac ports to make it do what i wanted to and not install
3:35 so many dependencies and stuff and then max was a friend of a friend of mine i think we showed fred
3:41 emmett maybe in common max um and he told me about this guy max who was doing this thing called homebrew
3:48 and then i went and looked at it and i was like oh it's written in this weird ruby thing which i
3:53 hadn't really had any experience over the time um and yeah i basically just ended up getting involved
3:59 like bit by bit like later on i guess the same year that humbrew got created i think max created it in
4:04 may and then i think i started getting involved in like september or something like that and was like
4:09 a maintainer before the end of the year um and yeah and i guess i've never managed to give up since so
4:17 still working on it to this day excellent yeah so you both mentioned linux i have a question
4:22 what was your first linux distribution
4:27 ah susa susa was mine i'm not sure what the first was for me it was maybe maybe fedora technically in
4:36 terms of like the first i ever installed but the one that held the the warmest part in my heart in
4:41 that era was gentoo uh just for like everything being built from source and i had this old pentium 233
4:51 basically my parents old machine that they were going to like throw out which by what year was it
4:57 like 2004 or 5 was pretty long in the tooth and i remember like getting an ubuntu install like sorry
5:03 i gentoo install where i wanted the base system to work like the machine had to compile continuously for
5:09 about two days to like get the machine set up and it was such a glorious waste of my time but i felt like
5:15 i sort of learned some stuff which ended up useful in the homebrew days through doing that yeah emerge
5:20 was my first my first package manager from gen so yeah that was i remember those days um yeah so
5:27 maybe just for uh just for clarity and and completeness what is homebrew like i i don't know
5:35 if you have the same definition of homebrew for both of you but like what is it why are we here and why
5:40 are we celebrating 15 years of its existence yeah well always my definition it's probably not accurate
5:47 anymore but you know um i'd say like it's the base of an awful lot of modern software nowadays
5:54 is is the truth of it i don't think either me or mike imagined that it would become so you know enormous
6:02 when uh when uh we were working on it back in 2010 or so but uh you know it's just a package manager i uh
6:10 i was very fond of package managers at the time and switched to mac in a fit of rage one one day
6:18 because uh i upgraded my linux kernel and uh my wi-fi stopped working and i was stuck at home by myself
6:25 mirror before smartphones so i didn't have the internet to uh help me fix it so i switched to mac
6:33 and then while i was there i discovered that uh the package managers weren't quite as good on mac
6:38 as they were on linux i was used to like trying out all the different linux distributions and things
6:43 so yeah i'm homebrew as a package manager but i think we took a lot of new ideas and ran with them
6:50 and that's part of the reason that it became so prevalent and uh it it it was the right toward the
6:59 right time in many ways so yeah it's the base of many people's stacks nowadays uh and makes it possible
7:06 to get development done what do you think mike yeah i mean the when i describe homebrew to non-technical
7:15 people i guess i go for like it's i basically try and get them to be like do you know what an app
7:20 store is and they're like yes and then if i can get them to be like do you know what open source
7:24 software is yes do you know what a terminal is yes well it's an app store for open source software
7:30 in your terminal right and that's maybe slightly reductive but like that feels like that's what it
7:34 has become in essence for pretty much everyone on mac like if they're installing open source on their mac
7:41 they're installing either the open source itself or the thing required to install the open source
7:45 they want if it's like npm whatever pro will be using homebrew together and yeah i i think you're
7:51 right about like the the time and the moment like i to me that the two things that felt linked in some
7:58 ways were because you know homebrew was one of the first big projects to kind of come out of github
8:04 and be very heavily community run and have huge numbers of community contributions i remember it was like
8:09 in the first few years that github would publish those reports homer was regularly in the like
8:13 the top charts for like most forked most contributed all this type of stuff and i think
8:21 that's basically what helped get it going in the direction that it went right is that because it was
8:26 github native before maybe any other big project ended up being that way uh i think that's yeah
8:33 it's a big part of it yeah like i have a small list that i've assembled over the years of the reasons
8:40 that i think homebrew was so successful but we can talk about that when when you like john yeah sure
8:44 well i think the next kind of area is just thinking about like the history and how how homebrew came to
8:50 be um what was i mean max this is probably mostly for you uh as you were the the creator what was the
8:57 the inspiration you know behind it and you also kind of mentioned like that it's gone a really long
9:04 way and it's it's become much more than you ever expected i'm curious to know as somebody who's had
9:08 some small open source projects what did you expect like you know obviously like i think it's pretty easy
9:14 to say that this went way past what you expected but like you had to be motivated in some way what were
9:18 you hoping would happen oh i was hoping this would happen
9:22 such amazing forethought yeah it's weird i had this feeling while i was building it that it was going
9:31 to be big and like that feeling didn't i didn't lose it and i didn't lose the confidence in it and so
9:36 i put my all into it and then it happened uh which was really bad for me i was 26 so uh i spent the
9:46 next like 10 years of my career thinking that everything i touched would turn to gold and uh
9:52 it doesn't it doesn't usually work that way i i definitely recommend not having a huge success when
9:58 you're young uh but um yeah i got over that in the end but yeah like i created it because i was working
10:05 at lost fan at the time and i was the lead developer of our apps team and we had six apps windows mac linux
10:11 android o phone and the blackberry app which uh was the worst and hardest to build but the unifying uh
10:18 machine to build all these things on was mac because partly because of apple's lock in for their
10:23 platforms but partly because you know it's unix underneath so you could run all the other different
10:27 sdks and systems and you could emulate windows uh so uh we were using a lot of open source tooling on
10:35 mac and like the package manager kept like coming up as a problem like mac ports was like gentle you
10:40 had to compile everything from source like mike was saying uh it insisted on building everything from
10:46 like uh glibc upwards even though the mac came with a bunch of uh open source libraries that are
10:54 usually that base that you don't have to build so i used to complain about how the package management
11:00 solutions were holding us back at the pub after work like on a relatively frequent basis to the
11:05 point that one of my co-workers got fed up with me and told me to just shut up and do something about
11:11 it and uh i went home and i was like well it would be kind of fun to work on something like that so i
11:17 started you know the that was the beginning i i called on like uh all the ideas i've had over the years
11:25 working on different open source projects i worked on this music player called america on linux for a
11:30 long time and uh the issues that everyone was always having were related to auto tools and getting
11:37 the dependencies in place and building from source and so i wanted to build something that i felt could
11:44 be a solution to those kinds of problems for people trying to build out open source work on open source
11:49 work in open source and do development in general uh so i wanted it to be a programmable package manager
11:57 so like all the formulas i uh called them uh were in ruby so that you know we're programmers i figured we
12:04 wanted a programming language in order to like build these complex and different pieces of software and
12:10 i picked ruby because it was very fashionable at the time and it came with the mac ruby on rails was
12:17 really getting uh a lot of attention at the time uh it's one of the reasons i was successful for sure
12:23 picking ruby because the ruby community loved that it was written in ruby and that you could use ruby
12:28 in order to uh work on homebrew um and then yeah like picking github was a big part of the success as
12:36 well for sure like github was relatively new right i think github had come out a couple of years before
12:41 that and we've been using github at last fm and i saw that uh it was a great place like i think the
12:46 alternatives for source forward right we we've been struggling with these terrible uh tools for doing
12:52 open source work online for a long time so yeah i picked all these things and put it out there i managed
12:58 to get some attention for it with uh replying to some people on twitter you know twitter used to be
13:04 a better place dare i say it where it was an awful lot of tech people so you could go on there and
13:13 just talk to tech people and see what they were doing and talk about like new tools and things and
13:17 so i got the attention of simon willison probably you've heard of him with uh tweet about homebrew and
13:23 then i got the attention of josh peak who was working at 37 signals at the time
13:27 ruby on rails uh was built by them and he tweeted out about how
13:33 he was going to install homebrew on his new mac that he was getting that had osx leopard and like
13:39 that was the moment where it really started to take off before that you know i was telling people
13:44 about it they were like oh so i'll tell call me call me when like anyone else is using it a typical
13:48 open source start you know like when you make some open source you do have to do a bunch of marketing
13:54 yourself in order to uh get some attention for it so yeah i was really motivated i really wanted it to
14:02 succeed and yeah like those first few years it was crazy uh we were dealing with a ton of uh issues
14:11 pull requests well pull requests when they turned up you know when we when homebrew initially was
14:15 created the pull request was not a thing nobody had figured out how to uh handle contributions on github
14:23 at this point they had forks and you could browse the four graphs so that's what we used to do right
14:30 like the uh the first person that i merged some of their contributions in that he didn't message me to
14:36 tell me that he was building these formula i just found his contributions in the ford graph and then
14:42 merged them in and then message him saying hey i merged your stuff in i hope that's okay because
14:48 that was how we had to do it at the time you know i'm waffling here and mike how about you where did
14:54 you uh how did you come across homebrew and kind of what got you involved and why did you why did you
15:01 decide to become a contributor yeah so i mean i guess i i used to hear a lot of people saying this back in
15:08 the day not as many people say it now is the idea with open source of like scratching your own itch
15:14 right and we've just heard max explain how homebrew came to be created from his scratching his own itch
15:21 right and it was the same with my contributions really where i heard about homebrew and i thought it
15:25 was cool when i started using it but i didn't have this idea of like oh i'm gonna get involved here
15:33 or i'm still gonna be doing this in 15 years or any of that sort of nonsense it was just like oh
15:40 like it's a package manager it's missing some stuff that i want so i'm gonna add some stuff
15:45 right um and every time i came across something that i needed i would go and submit a contribution and
15:51 then eventually the package manager itself didn't do some of the stuff that i wanted it to so
15:55 i submitted kind of changes to that and and whatever and i think it was i don't know again
16:00 as i mentioned before is that this all being on github there was something about it being on github
16:06 that just felt so much more accessible i think in the early days um the there was github and i guess
16:14 before pull request there was like an irc channel like i vaguely remember like you know instead of pull
16:19 requests i would like dm max on irc and be like here's a commit and then max would like
16:25 cherry pick that commit in and push it and tell me what it was pushed and stuff but yeah it just it
16:29 felt like the barrier to entry was very low um but i guess i guess the other thing that kind of captured
16:34 my interest with homebrew which again i thought was very different and very creative from the outset was
16:41 just the way that errors were handled you know most software most of the time particularly relatively
16:47 immature software generally operates from the assumption that nothing will ever go wrong and if it ever does go
16:53 wrong then like just throw your hands up in the air and just be like oh well you know like some programmer
16:58 come along and fix this right whereas i remember right from the outset homebrew was really impressively
17:06 well designed to give good error messages and even to stuff like the i've seen a bunch of other tools
17:12 copy this now but like the kind of brew doctor command which again was relatively early on but this idea of
17:18 being like hey like we have some ideas of stuff that could be wrong right like run this command and
17:23 these are not necessarily all problems that you need to fix but like if you are having problems
17:28 chances are it's going to be one of these you know um and yeah i just felt like that approach just as
17:35 what it it gelled with my brain and it sucked me in and it just felt so much more accessible and maybe even
17:44 scalable in hindsight than the models that a lot of other linux distros were having and other package
17:52 managers on mac that felt very much more like the linux distro model we're having
17:56 yeah like it reminds me it's just that from the start i knew i didn't want to do it all by myself and i had
18:04 faith in the users that if you give them the right messaging and docs and tooling that they can fix their
18:12 own problems and i really wanted to encourage that so you had the brew dog to come on the good error
18:17 messages um even the way like you cloned homebrew initially and that's where it built and so you had
18:24 a clone and you also could immediately start contributing there wasn't any like fancy build
18:30 steps or anything that was required you just opened up your editor and make changes and then push them
18:35 back to github those things i just really wanted everybody to feel like they could participate and
18:41 you know you've seen some projects do more of that since homebrew i think sort of introduced the idea
18:48 of doing it but still i think in general like you go to a readme for most open source projects and
18:53 there's no section saying hey how do you get started you know it tells you how to install it
18:59 but it doesn't say how you can get involved in the project i still think it's very important it's rare
19:06 so 15 years is a really long time um homebrew has been around for for a very very long time i was
19:14 hoping we could do a little bit of a you know a short timeline of the major milestones in the history of
19:20 homebrew so um starting from and i'll share my screen for this um let's see if i can do this
19:27 starting from here uh max this is probably for you um this is the first commit uh to homebrew
19:34 let me see if i can make it a little bit bigger
19:36 here we go so the first commit of homebrew um maybe you want to tell us about your state of mind when
19:45 you were doing this and any other kind of interesting uh anything you'd like to share
19:50 i'll leave it up to you yeah well um i still do this i don't start a project with code i start with
19:57 the readme and this was something i learned while working on kde i went to a kde conference and one
20:04 of the people that worked on cute which was the c plus plus toolkit that kde used uh still exists i
20:11 believe yeah i did a presentation where they talked about how they didn't start with code they started
20:16 by writing out what the api would be doing example code showing how they wanted the final api to be
20:25 and then they build the code around that so i started doing similar things but yeah here i was starting
20:30 with the remi so i spent more time on the remi initially before i started building any code out
20:36 and i was like how do i want this to be how do i how do i want it to feel and so yeah this this was
20:42 what i imagined it would be before i started writing any of the code and uh you know you can see a bunch
20:47 of the stuff in there that i wanted like the naming theme which homebrew is uh famous for which i did
20:54 because i knew that the world of package managers was so full of uh things that were difficult to
21:01 describe because the terminology was terrible of what does the word package even mean i don't think
21:08 anyone can really tell you the answer to that and so i wanted there to be a theme that allowed people to see uh analogies and metaphors in all of the uh the system and the code
21:23 uh so that it would bootstrap their ability to understand how to contribute or how to use it so
21:29 yeah you can see like the slash brewery there which you can no longer create that folder on macros thanks
21:36 to apple but at the time i think that was where i thought it would go eventually we changed that to use
21:40 a local right because it made so many things work better it was an important decision in the just works
21:48 aspect of homebrew which was important to the the way i designed it the way i thought about it
21:54 yeah like uh this is this is what i wrote with the typos and bad uh grammar throughout and this is where
22:03 it all this is where it all started um just on the theme of uh of homebrew i think that there's something
22:09 like in that era of ruby time i feel like every ruby project had some kind of an interesting theme
22:15 as well did you see that elsewhere was was it mostly like a or is that a ruby thing that am i
22:21 misremembering that or is that how it was no you're probably right i i probably don't remember it but i'm
22:26 sure that i was inspired to do the naming theme by other great projects and they just really suited what
22:31 i wanted to do i think my reasons for doing it was slightly different i think a lot of projects at the time
22:37 that had themes we're just doing it for the funsies while i was trying to make it useful and then the
22:44 vast majority of open source at the time was just like terrible naming and terrible if any theming
22:52 um you know it would be called what it did you know uh decrypt library dot lib
23:02 so now you know we've seen kind of where it began um in the past 15 years you know if we had to
23:08 highlight maybe three or four big inflection points in the project what would you say you know
23:12 maybe um were the biggest ones or the most the most significant uh to share
23:18 yeah it's a difficult one
23:24 i could i could suggest a few i think taps were the first that was one of your big creations max that
23:31 was probably one of your latter like big creations on the project where you just came out of nowhere
23:36 and again similarly like had this design and you're just like here's taps here's what it looks like
23:41 here's how it plugs together and i think that's been like before so taps are essentially like third
23:46 party repositories for homebrew if anyone who doesn't know and before we had taps you could only install
23:52 stuff in homebrew that had been added to the main homebrew repository essentially and i mean there were
23:57 other ways you could sort of hack around and stuff but i think taps are what blew homebrew from turning
24:05 from like just a project to being like an ecosystem because then you can have well anyone can now create
24:11 their packages in any way if you're a company you can have internal taps if you're um you know if you're
24:17 an open source project you can have your own tap if you're and even within homebrew itself like homebrew
24:21 now has a bunch of different taps and that's like homebrew cask that got merged into homebrew itself
24:27 started off as a tap all the linux stuff started off as a tap like it's just i think it just enabled
24:33 kind of the ecosystem to sort of flourish a bit more in that respect um i guess another one i might
24:40 suggest after that would be like my uh true love and joy is like bottles the binary packages that
24:46 uh because i think i implemented the first version of that and we're still using more or less a pretty
24:52 similar model today um and yeah and i guess over time we kind of moved from being almost from exclusively
25:00 built from source to predominantly built from source to mainly built from uh binary packages as it is kind
25:05 of today yeah can you say a little bit more about what bottles are and how they work
25:10 yeah i mean basically a bottle is just essentially we build everything from source the way we used to
25:16 uh on user's machine but now we do it on github actions runners uh we basically make a bottle which
25:25 is a complicated uh format of just a tarball of the thing that we installed again i feel like i learned a lot
25:32 i learned a lot from max in this era and in general that like often there's these really overtly uh
25:39 complex formats that happen on this stuff and often it's like well if you've got a directory of stuff
25:44 and you want to put the directory of stuff somewhere else can we just make a tarball of that directory
25:50 and extract it somewhere else is that good enough and the answer from the last 10 plus years or whatever
25:57 seems to be yes right and instead we could have made this complicated custom binary format and
26:02 made everything three percent more efficient or whatever but then the whole system becomes harder
26:07 to explain and understand and all the tooling you use is different yeah and whatever and i i don't i feel
26:12 like that's for me like a big part of homebrew as well has been trying to avoid that unnecessary
26:19 complexity and you know i think that's something i took from your initial design max and from
26:24 us working together over the years and has inspired me in homebrew and outside of homebrew as well
26:29 um yeah well it's good to hear because yeah like before that i i'd already recognized how complexity
26:37 can destroy productivity and the ability to iterate and it was uh it was a key thing you know one of the
26:42 things with the initial remi that we were just looking at was i wanted you to be able to use the unix
26:47 tools you already knew like rm and ls to uh query the package manager i didn't want it to need another
26:57 tool apart from when it was absolutely necessary and those sort of designs filtered through throughout
27:03 i think in open source in general there's often a a desire to over complex things um partly because
27:11 it makes us feel good hey i designed and built this really complex thing but it can hinder like i think
27:17 most of the most successful open source projects are simple enough that people get the gist really
27:22 quickly that inspires them to want to continue diving in and then get involved and open source
27:28 lives and dies by a community like it's one of the key aspects that enables it to exist and for it to
27:37 thrive so yeah and bottles yeah it was super super important it was interesting at the time right because
27:43 when i built homebrew initially i got a bunch of like negative feedback from people saying oh god you're
27:49 building from source i don't want to build from source just uh you should make binary packages
27:54 i'm like well yeah sure but who's who's going to pay for that because at the time there wasn't stuff
27:58 like github giving you free bandwidth and storage and compute in order to do these things as an open
28:05 source maintainer you had to figure out who was going to pay for it and i wasn't going to pay for it
28:11 yeah i mean the first version of our binary packages like the the evolution essentially went from being
28:17 built on my machine to being built on a vm on my machine then we did a kickstarter and then they were
28:23 built on some macs then built in a vm on those macs and then the macs ended up being cloud hosted and
28:28 you know like it's yeah it's i i think it's easy to look back at those days and be like why didn't
28:33 they build a binary package manager from the outset and forget that like what you mentioned already max that
28:39 pull request didn't even exist like travis ci and all those systems github actions none of these existed
28:44 you know even aws like you know it existed but you couldn't use it for building anything on macs
28:50 uh so yeah like i mean essentially you were like right well you somehow have to have build a bunch
28:55 of hardware yourself and build all this all the integrations and decide how that goes with the
28:59 workflow and and all this type of stuff and and even then like i think the homebrew workflow the
29:05 contribution workflow meant and means that like building binary packages is trickier than if
29:11 you're like a debian and you do a release of software every so often you can just be like okay
29:17 i maintainer has changed the thing we'll kick off a job the binary package is done when it's done you
29:22 know like it's much more complex when you have a user who submits a thing and you want it as soon as
29:27 the user merges the thing you want the binary packages to be ready for the user and all this type of
29:31 stuff so yeah it's it's interesting it's been a an interesting transition but a worthwhile one i think
29:37 something that i think is also kind of a milestone and you may you mentioned it a little bit mike
29:41 uh is casks so as a you know occasional contributor and a heavy user of homebrew i use casks all the
29:48 time and i really love the feature uh you mentioned that it was previously a tap and then it got merged in
29:53 i don't actually know the history or the backstory of how that happened but i think it was a pretty
29:59 important thing in the history of homebrew maybe you could share a little bit about how that came to
30:02 be yeah so i'm not entirely sure i think it was i can't remember the name of the guy i think it was
30:08 fins was his uh github handle but yeah someone set up this homebrew tap with enormous amounts of
30:15 custom ruby code and it was probably the the homebrew tap that had the most additional stuff in it with
30:21 this they came up with like a new dsl and this new concept called casks um for installing like binary
30:28 packages you know google chrome one password whatever it may be um and yeah and then after this project
30:34 been going on for a while like we'd kind of built closer and closer ties with that project and then it
30:39 was someone i think it was a a russian google summer code student called anastasia um and she basically
30:46 was involved with like merging the two repos and merging essentially all of the like the actual
30:51 like underlying like code logic for casks into into homebrew and then yeah like it everything was
30:59 pretty separate to begin with but then over time like casks and formula have been more and more kind
31:04 of like together and then now it's i think the nice thing that casks do is that it's you know it means
31:12 fewer people use them by quite a long way but it means if you want to just use homebrew to install
31:17 pretty much everything on your computer you can do that now which is between the mac app store and
31:23 casks and uh like humber itself and then there's the thing brew bundle which lets you kind of put them
31:29 all into a brew file and kind of have a somewhat decorative format and that supports installing things
31:33 from the mac app store as well so yeah so you can basically like i have essentially everything
31:38 my machine is being installed through homebrew in in some shape or form and it's yeah it again has it
31:45 broadened the ecosystem and wasn't was probably the biggest idea that came from outside of the core team
31:52 and got merged back in and other than we can talk about this later john uh like the linux sport as well
31:57 which had a similar kind of evolution yeah i have that on the on the agenda so any other big milestones
32:02 that you guys wanted to talk about i just want to talk about casks for a bit because uh i don't like
32:08 casks right uh i never like casks and uh it's very popular thing if you talk to anyone about homebrew
32:14 they're like oh yeah the first thing i do is cask install everything on my computer and then like
32:19 all the other tools as well i don't like casks uh but you know it was important that it wasn't all
32:24 about what i wanted and i always made sure that was the case and so we built brew like all the time and
32:32 mike was extremely instrumental in some of these pieces right i remember when you merged the sub
32:36 command support that wasn't something that i was thinking of but you added that so that people could
32:41 do brew dash whatever put it in the path and then that would be something you could execute via brew
32:46 and uh it was just always important that we allowed the community to experiment and see where they
32:53 wanted to take through and then if the community liked those things then yeah we would support them
32:58 or merge them in more fully so yeah it's important as an aspect of how to manage an open source project
33:04 but you allow that kind of thing i don't know exactly why i don't like them but yeah just a matter of
33:11 preference i think that's fair everybody deserves their their own preference um you reminded me there with
33:16 the external command stuff max it was also nice because we as you said it let the community go off and do
33:24 stuff um and there were a few times that people proposed kind of fairly massive changes to homebrew
33:28 which would require a lot of work and we were able to say hey we'll go off and uh go off and build this
33:35 in a tap and then if that gets some sort of you know adoption then we might consider emerging in later
33:42 and uh a little secret for any of the open source maintainers who might be watching like a lot of the
33:46 time in my head i was just thinking like there's no way you're actually going to do that but so one
33:51 of the the early successes of that i think before even casks which someone said like hey like we should
33:56 have descriptions for all the formula like a nice little textual description of what it is
34:01 um and it was like okay well that's going to require someone to go through and add lines to 5 000
34:06 packages describing them all i was like if you go off and do that and you get a description for
34:11 all the formula then we'll consider merging that back in thinking that person is not going to do it
34:16 no one's ever going to do this this is fine i've essentially told them to go away in a nicer way
34:20 and then you know this person reappeared in like you know four or six months or something and was like
34:23 cool yeah i'm done i'm done now i've we've got descriptions for all of them and then we merged it
34:28 back in and then with with a lot of these changes it's then once you got that thing it's much easier to
34:34 say hey every new formula we add has to have a description now and it felt like a big leveling
34:39 up for the project but you know it felt like i guess as max said earlier like you have to be
34:44 um you know i think it's good having a vision of what you like and what you don't and and even like
34:50 max not liking costs i think like that's a good example of like you know we we all have our own
34:56 preferences and our design decisions and whatever and we can't expect them to be the same but at the same
35:01 time max was not trying to ensure that casks could not be created by someone else and the same with the
35:06 description stuff i thought this was not going to happen but i wanted to like leave the door open
35:10 that like well if someone goes and essentially does all the work for free then yeah we can we
35:15 can include descriptions why not yeah i deliberately didn't put descriptions in the initial formula
35:22 and i remember like fighting people to keep them out so it's all objective everyone like i have a different
35:28 description we'll end up with edit wars and this information is easy to google etc but like uh
35:34 yeah it's very useful metadata for sure um
35:38 with like package x which uh i recently built right which is an attempt at something new in the
35:45 package management space didn't add descriptions as well and i was like oh we use ai to generate
35:50 descriptions um which mostly worked but then for some packages it was just abysmal and we actually got a
35:57 lot of backlash to that so uh it's um a difficult one sometimes figuring out what you want to support
36:04 because when you add something like that yeah then that means that every package from then on needs
36:09 a description so you're increasing the burden but that was very minor thing should have been in there
36:14 from the start probably so i kind of wanted to shift uh gears a little bit and talk about apple
36:20 um so it's pretty intertwined like homebrew is pretty uh there is a linux support and some other stuff but
36:28 um you know max you mentioned when you came over to the apple ecosystem that there was no good package
36:35 manager you weren't very satisfied and i think that homebrew actually has really really improved the
36:40 experience for developers on apple devices and apple changes relatively quickly uh you know the operating
36:46 systems and whatnot like they're they're always advancing their underlying architecture um so i wanted to
36:51 talk about maybe two or three things there one is the uh the move to a unix like posix based
36:58 operating system right uh with mac os 10. uh the other is the constant march of progress with apple
37:05 operating systems deprecating things changing the rules on the platform and then also their architecture going from ppc to intel now apple silicon those have all been major changes
37:13 relevant to homebrew and i was wondering if you know either you had commentary about that or
37:21 experiences from you know using those things and going through them yourselves yeah i'll quickly chime
37:27 in then pass to mike because i think he has much more experience in uh the various important changes
37:32 especially the uh apple silicon one more recently but um i think at this point honestly
37:39 apple cannot change some things about mac os because of homebrew
37:43 they they would risk um enormous problems for the global developer ecosystem if they did
37:53 but yeah earlier days it was more problematic but i've been very good about binary compatibility
37:59 uh in their libraries we only had a few issues um every time they do an upgrade
38:05 in general but yeah mike has uh more experience yeah i mean i guess the early one is you
38:12 initially released homebrew on leopards and then it was the snow leopards i think leopard had some
38:18 x64 support right and that but it was like not really and then snow leopard went a bit more all in on
38:25 x86 64. like so that transition was like i think the first and a relatively easy one um shout out to misty
38:35 de mayo a former homebrew maintainer plc member uh who had a thing called tiger brew which like back
38:42 ported homebrew to run on like pvc chips i i believe is still kind of working on those ancient apple devices
38:48 today um and then yeah i guess you mentioned john more recently the kind of apple silicon transition
38:54 i i mean i think like you know it's no secret that every new version of mac os tends to break some
39:00 stuff in homebrew or either mac os or xcode or whatever um but like you know i i think compared
39:08 to the earlier days we definitely have a private but decent communication channel with apple when stuff
39:15 really needs to be kind of addressed and particularly the apple silicon transition was made much easier
39:21 by apple like proactively essentially as soon as the keynote was done where they announced apple
39:26 silicon like before there was any hardware in anyone's hands they reached out to us proactively
39:30 and we're saying like hey how can we get the right apple silicon in the hands of the right people
39:35 in the right place and whatever um and yeah and i think it's it's funny because i think there's
39:40 definitely a bunch of two-way learning that's gone on i think homebrew's learned like how to identify apple
39:47 and how to work with them and figure out like what that looks like but i i think apple's also like
39:52 been getting better with open source and learning that like you know sending hardware to homebrew people
39:59 may involve sending it to like a this is not even a a uh fictional example but like some distant scottish
40:07 island um and like you can sometimes hear the confusion the apple people's voices that it's like
40:13 you're not just all in this office somewhere in the world you're like scattered all around like what's
40:20 the legal entity we should attach this to oh there isn't one oh uh this is going to be tricky uh but you
40:25 know in the end i think we've got something that i think work brutes or homebrew has been tremendously
40:31 beneficial to apple over the years and i think you know obviously if we didn't love our apple devices
40:38 like homebrew probably wouldn't exist in the first place because we would have just stayed on linux
40:43 doing linuxy things with linux package managers which were good enough that homebrew would probably have
40:48 not been created yeah and speaking of linux and also other platforms i know that homebrew
40:55 has support for linux there's linux brew right and there's also uh what is it uh the windows uh wsl
41:04 some people can use homebrew there so what's the project's position now on other operating systems is
41:11 it like officially supported by the project how how do you approach that yeah so linux is officially
41:17 supported with homebrew now and it's kind of fully integrated in terms of the package manager itself
41:23 and all the formula and homebrew core are run on you know this specific formula that might be mac only
41:30 or linux only or whatever but essentially everything we are aiming to have feature parity version parity
41:36 and everything like that and we have maintainers who run exclusively on linux and maintainers who run
41:40 exclusively on uh mac but mac is still our predominant uh ecosystem if it feels like an apple
41:48 level uh perhaps slightly silly discussion but um there's been a lot of back and forth on pull
41:55 requests and internally about like well how do we communicate that because homebrew is still used way
42:00 more on mac os than on linux um so on the home page we have you know homebrew the missing package manager
42:06 for mac os or linux and the or linux is parenthesized to sort of indicate that we do support linux but it's
42:13 like you know not quite as popular or you can look at the numbers i think linux usage is about maybe 10 or
42:20 something of mac os usage based on analytics and also like maybe interesting maybe boring uh like mac os
42:28 is like predominantly people running on their actual match and like i think it's like 10 20 or something
42:33 it is running on like ci environments whereas linux is like 50 actual users 50 ci so i think that's
42:39 again where the linux support is kind of kind of useful like even for us in homebrew ourselves like
42:43 having the ability to like bump mac os packages and like run style checks and all this type of stuff on
42:49 linux runners because i guess despite our conversation earlier where you know back in the
42:54 day when there's no mac os versions mac os runners available even nowadays like the mac os
43:01 compute per minute price is dramatically higher on any platform than the linux uh compute price so
43:08 essentially if you can farm a job out to a linux system instead of a mac system it makes life much
43:13 easier but again like we said earlier like the linux stuff started with a fork of homebrew
43:19 called linux brew over the package manager itself and then it had its own version of homebrew core
43:24 where all the formula are called linux brew core and over time the package manager in uh the main
43:31 package manager repo homebrew we call it was kind of made to be more and more linux compatible and
43:37 the eventually like the homebrew core became linux compatible as well and a little shout out to the
43:43 past max mentioned he was working when we were in london we were both working on qt applications actually
43:49 and yeah and i felt like from my cross-platform qt days i learned a lot about how to make the uh
43:55 the package manager part of homebrew not just a mess of like if mac if linux if mac if linux like all
44:02 over the place which is what kind of naive cross-platform porting tends to look like and be very painful
44:08 to deal with max so you you had mentioned that uh you know you don't recommend having a big success early
44:15 in your career uh when you're when you're young i'd love to hear from you um kind of how homebrew has
44:21 impacted you professionally and what that's done to your kind of trajectory as an engineer and as i know
44:28 you're an entrepreneur like what are you up to now where are they like where are they now and like
44:33 how did homebrew kind of put you in that path yeah well uh homebrew definitely impacted everything about
44:41 my life very much that's for sure hopefully for the better yeah well definitely like um i can't talk
44:49 fondly enough about how open source has given me so many opportunities that i wouldn't have otherwise
44:56 had but uh you know let's start with a personal level like suddenly i became very arrogant for a
45:01 while about it because uh you know it was the biggest project on github and it was on my username
45:07 slash mxcl slash homebrew um and uh you know i'd meet people and they i remember one conversation
45:16 in like 2014-ish maybe 2013 and i was sitting at a bar with some other devs and this guy says oh have
45:24 you heard of homebrew this package manager it's really cool and i just started laughing and then
45:29 everyone else around me was like what did he ask and i said oh he asked me if i'd heard of homebrew
45:34 and they were like and they talked about how i was the creator i've always been somewhat humble in
45:40 that respect i don't usually bring it up because i find that you get a bit of hero worship what you
45:46 probably find similarly mike and uh it changes how people act with you so i've had this
45:54 interesting sort of life where um among like geeks i'm famous but among everyone else fortunately i am
46:02 not i don't i don't have to be uh a celebrity in uh in most circles and uh you know it's definitely
46:11 got me career opportunities and you know obviously famously didn't get into google but
46:17 not getting into google got me an awful lot of other opportunities and yeah i think the worst thing was
46:27 definitely that after homebrew nothing i could work on seemed successful unless it had you know
46:35 a frac some some fraction at least of homebrew's success so i abandoned a lot of open source projects
46:43 i probably shouldn't have because homebrew was successful very quickly and i turned down a lot of
46:49 interesting startups that i could have worked on because they just didn't seem cool enough
46:53 um but yeah still always done open source i've got a couple of other big projects on my uh
47:00 github and uh missed out on some interesting airdrops because mike decided to deprecate the
47:09 original homebrew repo at some point you owe me uh some token for that like yeah um it was a couple of years ago
47:17 years ago that i founded t protocol which is what i'm doing now and uh it was based on my career in
47:23 in work and open source and like the the fact that nothing but open source ever really felt worth
47:30 doing to me and over the last 15 years uh open source has really gone from being something that was
47:37 kind of derided you know like it you could see how open source was becoming successful 15 years ago when
47:44 homebrew was created but it's still the case that if you wanted a real job then no one was using open
47:49 source and they thought you were a hipster or an idiot for wanting to use it so now we're like the idea
47:55 of building something without an open source base is preposterous and um yeah in those 15 years i think
48:02 it all kind of took us by surprise how much open source became the bedrock of all modern software
48:09 development so a couple years ago i was looking for funding solutions for open source once again
48:15 and i didn't find anything that seemed like it was going to work for me i'd already tried a few things
48:20 so vanity protocol to try and fix it by changing that narrative it's a crypto project which obviously
48:26 a lot of people are skeptical of but we're doing very well so the idea is we're trying to make open
48:34 source sustainable by giving free crypto token rewards to open source developers and we have almost 1.5
48:41 million people who've signed up for our testnet so far at 15 000 open source projects some of them uh
48:47 very impactful ones so yeah we're launching it properly later this year and the idea is to make
48:53 it so you know like my homebrew story was that i quit my job to work on homebrew full-time and if i hadn't
48:57 done that i don't think it would exist quite honestly and it's not really something we can
49:04 expect the people who are building the foundation of the internet to do on a regular basis we can't
49:10 rely on the software industry and all the software that we build that's so important nowadays to
49:16 exist when we were expecting people to somehow try and make these sacrifices i couldn't do that now
49:22 at 26 you know i was single didn't have kids it was fine but so yeah we're trying to change that
49:29 t protocol makes it so that you don't think of open source as a charity which is how everyone
49:35 else goes about it you're begging for donations changing into like a value ecosystem where
49:40 you're you're participating in the value of open source
49:48 excellent yeah i mean i can only imagine that homebrew has had a major a major impact on your
49:53 entire life and i can relate to what you were saying about you know when you have a big success
49:58 it's very difficult to try to do another project and have it be successful like i've i've kind of
50:04 developed this muscle of don't try to do the same thing twice um where if something happens to be a
50:10 home run it's a combination of good good work you know forethought but also a healthy dose of luck
50:16 and you can't expect the luck to be there every time um but yeah i definitely i definitely can
50:21 relate to some of the things that you were saying right like um yeah mike how about you how has homebrew
50:26 impacted your career and uh you know where are you now what are you up to these days yeah i mean
50:32 similarly to max like i've had a bunch of doors that got opened probably only due to homebrew right so
50:38 i started working on homebrew and ruby and you know i dabbled very briefly one university course back in
50:45 the day but um you know homebrew is what kind of got me doing ruby and then i've definitely found like
50:52 there was a few as homebrew became more successful uh when i changed jobs and i had like homebrew maintainer
50:58 on my resume like that seemed to be the only thing that people would care about and that got me through the
51:04 door and got me kind of speaking to people and i i applied for github three times before i got a job
51:09 there but you know i went from essentially being a nobody who uh like just a random member of the tech
51:16 industry to kind of being able to get a chat with one of the github founders back in whatever it was
51:21 2012 or 13 or um and yeah and i think that door was opened exclusively because of homebrew and i think
51:28 yeah i spent 10 years at github i left there last year and i think i learned a lot about making homebrew
51:33 better from github and a lot about making github better from homebrew and there's probably some
51:38 stuff i'm not going to go into right now that hope that github does differently as a company
51:43 uh and as a software project because of homebrew right and if homebrew didn't do it first then
51:47 github probably wouldn't be doing it um but yeah nowadays i guess kind of like max i i've also almost
51:54 like had a like thinking about what does this look like going forward and like what what do we do here
52:02 and i co-founded workbrew with yourself john and our co-founder vanessa who's not on the call and
52:09 we're basically doing i guess it's interesting how like max and i are both kind of riffed off the the
52:14 homebrew idea in different directions but basically what we're doing is building uh commercial support for
52:19 homebrew and what that looks like today is providing um better support for the stuff that homebrew doesn't
52:25 really want to do that the volunteers are not interested in like making homebrew work really
52:29 well with mdms in environments where users don't have administrators and tightly integrating with kind
52:36 of security protocols cvd disclosures and stuff like that but moving forward we're basically making
52:40 homebrew better for engineers using it as well and security professionals too and so yeah we're basically
52:46 working on making homebrew better but also providing a kind of commercial layer on top of homebrew uh
52:51 to make that better for the people who are using it i love what you guys are doing and i wish you
52:57 all the success with it i think it makes a ton of sense honestly i remember in um i think it was 2014
53:04 i got a call from some uh san francisco vc who was like how can we make homebrew into something
53:11 commercial because they they were looking at npm right which had just just done that and at the
53:16 time i was like i just don't see how we can do it because i didn't have the right mindset and also
53:20 because it was just such a big open source project i didn't understand how we could like say hey community
53:25 um it was open source but now it's not uh it's kind of something else so what you're doing is the
53:32 right way to go about it i wish i thought of it then might have been a very different world for
53:37 me and all of us but uh you know building on top adding extra services on top i think it's only the
53:42 last few years that commercial open source really people figured out how to do that and like i think
53:48 there's a lot of good opportunities in that space well i mean the funny thing is john had been probably
53:53 from a similar sort of era trying to convince me to that this was a good idea and i felt the same as you
53:58 that like well i don't really see how we can do this but for me it was like two things kind of came
54:03 along so one thing was the um i guess like homebrew actually having an established government structure
54:10 governance structure and its own funding and a non-profit and a board and all this type of stuff
54:15 like and that gave me the separation that i feel more comfortable with where it's like okay now it work
54:21 brew even if we wanted to try and somehow take over the open source project we can't do that because
54:26 it's its own thing and it has voting and everything like that right so that that feels a lot more
54:30 comfortable to me um but then also on on our side i guess having seen over the years something i
54:36 actually learned a couple of companies before homebrew is it was this company kda i used to work on again
54:40 doing qt consulting um and i was really excited because they were the biggest kitty contributors in
54:45 terms of a company that was out there um and i was like great i'm gonna get to work on all this cool
54:50 exciting open source stuff and get paid to do it and then i learned pretty quickly like oh no no if
54:55 you're getting paid to do the open source stuff you're getting paid to do the boring things right
54:59 because all the fun stuff people will be happy for free in their spare time and it's and i think one
55:03 of the projects was something like uh k office the kind of open office alternative and it was like
55:08 working on a compliance suite of like okay here's the like essentially 2 000 ways in which k office
55:15 doesn't render things the same as words just go through and like fix them all right and no one
55:19 wanted to spend the readings and weekends doing that so we were paid to do that right and i feel
55:24 like it's a similar model to what we're doing with workbrew essentially where there's a bunch of stuff
55:29 that i've seen over the years like humber maintainers are like ah i don't really like this is only that
55:34 you only use this in lockdown environments where security and compliance concerns are really important
55:39 like that's not me that's not my personal laptop like why do i want to spend like a day working on this
55:44 and so it doesn't get done and instead of saying no to people we can now say okay well like instead of
55:51 having homebrew solve all these problems for you which only apply really to companies that kind of
55:55 have these type of problems rather than individuals working at home then this workbook can solve them
55:59 for you instead so yeah um so just to wrap up we have a couple a couple minutes left but i i really
56:07 wanted to uh you know kind of see if there's any advice that you have for people who want to be
56:13 open source you know open source developers like you um if you were to go back to when you first got
56:20 involved with homebrew you know what advice would you give yourself at that time now that you've got
56:25 basically 15 years behind you
56:29 well i'll go first um so i always say that the most important thing is it's something you need
56:37 like in in general when you're working with you know you've got to be working on something else and
56:43 then as you're working on that other thing you're like oh this this thing doesn't exist or this tool
56:49 isn't what i need it to be i need it to be able to do something else and so if you're going to start a
56:54 new project like it's one of those things but if it's the latter you might find there's a tool that
57:00 exists and you can start contributing to that like you know that's a lot of open source people got
57:05 involved that way by finding that it had to be them who made that feature and uh if you have great uh
57:13 core maintainers like mike who encourage you to go do that partly because he's lazy partly because he's
57:18 made sure that over the years uh the ability for you to like bolt in your own functionality is easy
57:24 then uh yeah it's going to be something you're going to be passionate about if you're just doing
57:28 it because you think it's going to get you a good resume then you'll find after a few weeks you're not
57:33 doing it yeah i'd echo what max said like we actually have that in one of our like comfort
57:40 maintainer guidelines that like if you find yourself in a situation where you're no longer using homebrew
57:45 day-to-day like you're doing your day job on linux and all your personal machines are on linux and you
57:50 just don't use homework there or whatever it may be then we recommend people step down because at that
57:55 point not only like is it not that interesting to you but also like at that point um you know you're
58:02 probably not going to do the best decisions for the project because you don't care personally
58:06 and yeah like i i don't know like i think it's for me thinking back 15 years like i don't know i've just
58:12 i've learned a lot along the way that i don't know if i would have learned if someone told me because i'm
58:17 super stubborn like that um but yeah i guess for me the biggest transition has been
58:21 just having more boundaries and being better at like dealing with people being entitled or
58:28 like i no longer have like homebrew slack or notifications or whatever on my personal phone
58:34 and stuff like that so it's yeah i guess that has both helped kind of let other people pick up the slack
58:41 and like share the load a bit more but it also just stops random open source people who are 99.9 of
58:50 the time lovely and sometimes less than lovely uh from intruding on my personal life and my personal
58:57 space and when i'm having time with my wife and kids and stuff like that yeah well yeah mike was always a
59:02 lot better about that than i was you know he knew how to keep the rest of his life separate and i you know
59:12 that's why i burned out while uh mike kept going for sure so yeah something i can learn i think
59:22 we all can um well thank you both for for joining us um this has been awesome uh congratulations to both
59:29 of you also 15 years since homebrew was created and i think um thank you to everybody in the community
59:36 who uses homebrew who contributes to homebrew i think that all of the you know everybody who's got
59:40 commits in there whether it's you know features or documentation or really anything to help out with
59:46 the project you know you deserve to celebrate as well um thank you max uh thank you mike for joining uh
59:52 have a great day everybody bye bye thanks

Mike McQuaid

Homebrew Turns 15