Homebrew Unleashed: Diving into the Fast and Efficient Packaging Process

13 March 2024

Interviewed by Ruby Rogues

Mike McQuaid is the CTO and cofounder at Workbrew. They dive into the world of Homebrew, an open-source package manager for macOS and Linux. They explore the history and development of Homebrew, from its origins in the Ruby community to its evolution into a widely-used tool for installing and managing software.

Show transcript

0:00 hey welcome back to another episode of the ruby rogues podcast this week i'm talking to mike
0:11 mcquade mike is one of the maintainers of homebrew uh if you have a mac you know and if you don't
0:19 have a mac then we'll we'll get into what it is but yeah cool stuff uh calling in from scotland
0:27 is that what you said yep edinburgh and scotland very cool um i have some scottish in me i have a
0:34 whole bunch of other european countries in me too but anyway uh do you want to just jump in and let
0:40 us know what else people ought to know about you before we get rolling hey folks this is charles
0:45 maxwood i've been talking to a whole bunch of people that want to update their resume and find a better
0:50 job and i figure well why not just share my resume so you if you go to topendevs.com slash resume
0:56 enter your name and email address then you'll get a copy of the resume that i use that i've used
1:02 through freelancing through most of my career as i've kind of refined it and tweaked it to get me
1:08 the jobs that i want uh like i said topendevs.com slash resume will get you that and you can just kind
1:15 of use the formatting it comes in word and pages formats and you can just fill it in from there
1:20 sure so yeah as uh charles honestly said i'm mike mcquade uh he said my name perfectly which is not
1:29 always something you can rely on which is very nice and so i'm currently working on uh homebrew in my
1:37 you know volunteer life spare time sort of stuff worked on it for 15 years since 2009
1:44 2009 i got involved i think i was the third person to get a commit bit on the project and i'm
1:49 oh wow i'm still standing after that time and i'm like the project leader which basically means i'm
1:53 sort of an elected like figurehead slash tech lead etc um so yeah so in my kind of outside of my homebrew
2:01 life work-wise and i'm working right now started a company last year with some ex-github people
2:06 called workbrew where we're working on commercializing some stuff around homebrew and
2:12 basically building the features that kind of small medium large companies need that homebrew does not
2:17 have so that they can work in the way they want to work while homebrew plays nicely with that and then
2:22 before i was working here i worked at github for 10 years and i left as like a principal engineer
2:27 last year and i was bounced all over the place there but particularly i guess focused on like
2:32 developer experience stuff and open source and scratching my own itches basically
2:36 very cool yeah i've uh known a few people who've worked at uh github i also remember way back in
2:44 the day when chris and tom put their heads together and started the thing as kind of a project in a
2:50 restaurant or something and yeah it's kind of funny to look at where it all wound up with
2:55 microsoft buy in them and everything else um let's just kind of start with the fundamentals here
3:01 i did kind of mention that if you have a mac you probably know what homebrew is but do you want to
3:06 people a rundown of what it is sure so homebrew is a open source package manager for mac os primarily
3:13 and also increasingly linux as well interestingly oh really could talk a little bit more about that
3:17 later because that's sometimes confusing to particularly mac people um but yeah so basically
3:22 if you you write brew which is the kind of shorthand brew install and then pretty much any
3:27 application you would want and you can install that on your mac um so it's i believe i then kind of
3:34 we have some degree of evidence that you know about 30 million people and or bots make use of homebrew
3:41 and it's yeah in our opinion like the best way of installing open source software on your mac well
3:47 you may not know as well with homebrew it started off just being like a building everything from source
3:51 package manager for like open source tools wget curl whatever whereas nowadays you can also install
3:57 like upstream binaries so like say like chrome which i i'm a diehard safari guy so i'm streaming in chrome
4:03 today because i had my usual sadness of learning the website didn't support safari so i fired up chrome
4:09 but i i install chrome by just typing brew install google chrome and then it sticks everything in my
4:14 applications folder sets it all up and and all that type of things and there's a long tail of like
4:18 ecosystems where you can have like brew bundle is another favorite of mine where you can have like
4:23 declarative ways of telling homebrew what to do and all this type of stuff so yeah that's basically
4:29 and my uh my non-tech explanation of homebrew is like a app store for open source in your terminal
4:36 right yeah you just brew install whatever you want and off it runs and does its thing
4:44 in fact a lot of tutorials if you need a particular piece of the puzzle a lot of the gems and things
4:52 that i use in ruby it'll it'll basically say go brew install your blah blah blah right and then
4:57 and then you can run from there so yeah it's definitely something that i use and i reach for
5:03 and i don't even think anymore that oh gee somebody actually made this right it's just part of life so
5:09 anyway um so yeah so i'm kind of curious uh as we kind of get into this a little bit and there's a
5:17 lot to talk about right because i do want to get into workbrew and what what that's all about but
5:21 um before we do that uh for those of us who have used homebrew for a long time and are excited to see
5:29 um you know maybe where it came from and what all is entailed in it um where did it come from
5:36 you said you weren't in it from necessarily the beginning but you got to commit yeah pretty early
5:41 so yeah what's the story there i think i got involved about you know maybe started using it four or five
5:47 months and started contributing pretty short after when it was first created it was made in 2009
5:52 by a guy called max howell uh who was in london i think he was working at last fm at the time
5:58 it's a blast from the past that probably still exists but not as widely used as it once was
6:04 um so he had i have kids younger than homebrew yeah i'm sure yeah i i also have kids younger than
6:12 homebrew it's and it's distressing when you meet kids who are the same age as homebrew and how articulate
6:16 that they are um so he basically had played around with all the package managers on the mac at the time
6:24 you know there's mac ports and think and uh oh yeah i remember those yeah i can't remember what the
6:29 other one was um but yeah yeah mac ports is probably the dominant one um and he didn't have a huge amount
6:36 of like good time using mac ports felt that it kind of did too much and he didn't like the approach
6:42 so he built homebrew which was this kind of beer themed package manager um and yeah and the other
6:49 still called casks yeah the packages are still called like formulae and casks and it still got all the
6:55 the beer themes all over the place but one of the interesting things he did as well is it was you know
6:59 this was i can't remember you probably know charles actually what year github opened to the public
7:04 but certainly this was in 2009 so github was like relatively new i'm pretty sure when homebrew first
7:10 got created the pull request didn't exist yet and but max from the outset had decided like hey i want
7:17 to make this thing but i can't maintain thousands of packages all by myself so i'm going to have to
7:22 kind of open this up and make it a community effort to do that and so yeah so that's basically what
7:28 happened and more and more people kind of particularly in the ruby community like i think the ruby community
7:32 was one of the early adopters of homebrew because homebrew itself was written and it still is
7:36 primarily within ruby and uh yeah basically just it kind of grew more people got involved more people
7:42 contributed and then you know people who contributed lots and end up doing a bunch of reviews and stuff
7:47 like that i guess folks like me ended up kind of becoming maintainers and then the projects just sort
7:53 of grown and grown over the time so yeah it went from originally like just max in 2009 to nowadays
7:58 we've got i think 30 something maintainers we probably had like wow 50 total all time and then we've had
8:04 like over 10 000 people contribute to some part of the project at least once very cool so
8:13 i have so many questions um so one of them is is there there are kind of a bunch of different pieces to
8:23 this right you've got the command line interface and then you've got the hey go grab the thing and how
8:29 do i know how to build the thing and how do i know where to put the thing and how right and so um
8:34 let's just start with the command line interface so um what what tools did you start out working with
8:42 and and has that changed over the last what 13 14 years yeah yeah so homebrew was written originally
8:51 exclusively in ruby pretty much um and it's still primarily ruby like uh the the irony in homebrew land
9:01 compared to maybe other bits of ruby where because we're like a cli and because we're we are installing
9:07 a bunch of stuff related to compilers and things but we can't rely on them necessarily being there
9:10 and we haven't wanted to ship native binaries for the application itself when we want to make things
9:16 faster in homebrew often they are ported from ruby to bash so an increasing amount of wow they're kind
9:23 of very performance intensive bits of homebrew like the bits where like you want to be able to run a
9:28 command and have a response instantly or you want to be able to have a command that you can run every
9:33 time you open your shell and then not have too much overhead so some of that more of that stuff is
9:39 kind of ported to like bash and like the updaters and bash and because homebrew installs its own ruby
9:45 nowadays because the you know we originally used just the whatever ruby ship with mac os and that
9:50 worked all right for a pretty long time but i think last year we were speaking to our kind of apple
9:55 contacts and they'd theoretically deprecated ruby i think a few years ago in the system and we said hey
10:00 can we get a newer one i think it's still there it is still there but it's a 2.6 yeah i was gonna say
10:06 it's not a current version yeah pretty ancient and so yeah so we we ended up we we were rolling our
10:12 own for really old mac os versions and for linux anyway so we just like well we can roll our own for
10:17 for everyone now and just make it mandatory right and i think it's ruby 3.1 or 2 or something like
10:23 that uh-huh nowadays and yeah and that's a much nicer experience for us and makes everything a bit
10:28 faster and snappier and stuff so yeah so the main package manager is all in ruby and then
10:31 we have i guess two main repositories so there's like homebrew brew we would call it which is like
10:37 yard.com slash homebrew slash brew that's the package manager and that's like that's my personal like
10:43 favorite happy place to live and i kind of review pretty much every pr that is in that repo and and
10:50 then there's like homebrew slash homebrew core and so that's like the core formula like basically
10:56 everything in there has to be open source and there's like 6 000 packages or whatever in there
11:00 and that again is i think one of the reasons why ruby was such a good pick originally like is that
11:05 the like the the files that kind of describe how to install something or known as formula and
11:12 and again going down the beer theme and that it's like a really nice ruby dsl that kind of
11:18 is i would say fairly if you're if you kind of have a basic understanding of almost like you know
11:24 configure make make install like how unix programs are generally built like it's you can read through
11:29 that and it's pretty obvious how to get started and if you've ever had the this episode is brought
11:35 to you by spreaker the platform responsible for a rapidly spreading condition known as podcast brain
11:40 symptoms include buying microphones you don't need explaining rss feeds to confused relatives
11:45 and saying things like sorry i can't talk right now i'm editing audio if this sounds familiar you're
11:52 probably already a podcaster the good news is spreaker makes the whole process simple you record your show
11:57 upload it once and spreaker distributes it everywhere people listen apple podcasts spotify
12:03 and about a dozen apps your cousin swears are the next big thing even better spreaker helps you
12:08 monetize your show with ads meaning your podcast might someday pay for well more microphones start your
12:15 show today at spreaker.com spreaker because if you're going to talk to yourself for an hour you might as
12:21 well publish it uh misfortune of having to try and package like rpms or like debian packages or whatever
12:29 which are very very robust arguably better than homebrew once they're there but like the actual just
12:35 process of going from nothing to something that works is very time consuming in my experience um whereas in
12:42 homebrew you can just uh you know like make a fairly short little thing and then have yourself up and
12:49 running installing applications in a pretty quick time and and i think that's why it's grown and stayed
12:54 big because it's quite easy for people to contribute and modify and change things and it's just you know
12:59 you can do it all on github nowadays right so when i was in college i w i was a systems administrator on
13:07 mostly red hat systems and so i've i've written some fugly bash scripts to do all kinds of things and
13:15 and i've also done rpm packaging and you know later on i yeah there are a few things that i've had to
13:21 fiddle with and maybe put into a debian package and yeah that is not a fun process
13:26 and then you just kind of have to put it out there and then pull it in with your package manager or with
13:31 some utility that can pull it in and unpack it and do whatever it does and oh man yeah homebrew's
13:40 pretty nice that way uh also done plenty of stuff with make which is it's a powerful utility but it's
13:46 a little arcane sometimes and so yeah i i definitely feel you there so um
13:53 um yeah so it's all written in ruby is it pretty is it pretty simple or there are a lot of edge cases
14:01 you have to handle or is that mostly in the packages or the formula or whatever yeah a bit of both like
14:06 i mean there's i think with with stuff like homebrew is like a really long tail so i guess with the stuff
14:12 that max created in the first place um you know i can't remember how many packages he's almost like
14:17 shipped with to begin with but um it's one of those things where if you have a hundred packages
14:23 everything seems pretty simple if you have a thousand things start to get a bit more complex
14:26 and then when you start getting up to like five six thousand it's like it gets really messy all the
14:31 weird and wonderful things that things do and as a packager of software you have this kind of weird
14:38 decision framework where there's often the way the the people who write their software originally i guess
14:45 we would refer to them as upstream and the way they want to release their software is not necessarily
14:51 the way your users want to consume the software or the way you want to be keeping it up to date and
14:55 stuff so you're trying to sort of balance these things and yeah i mean the thing i've enjoyed about
15:01 it personally is like i've always been a bit of a kind of generalist software ride so like with ruby
15:06 particularly when i was doing a lot more of the packaging stuff than i do probably today you know in
15:10 a given week you could be writing ruby c c plus plus haskell like having to play around with like
15:16 10 20 different languages and frameworks and you kind of develop a sense of how this stuff plays
15:22 nicely together and what the what the edge cases are and what you need to watch out for and everything
15:27 like that but yeah i mean software is a a wily beast uh and packaging it's no less so
15:36 interesting so um yeah so what are your dependencies i know that you have to have the xcode build tools
15:44 installed does it yes in the rest of them for me yeah basically that's you on a mac like that's all
15:51 you need you just need to have the xcode uh like utilities installed um and then even then like we've
15:58 we've sort of tried in the past like sometimes to make them not required but it's it's pretty hard
16:03 given like the state of right things in apple land to to go completely without that and you know every
16:09 few years apple introduced some new requirement which requires using a cli that only they
16:14 provide and there's no open source version and we're kind of back to square one there so
16:18 yeah basically just the xcode command line tools are the only dependency and then if you're running
16:23 hobo and linux there's a few other bits and pieces there but and the big one i mentioned before like
16:28 ruby like we we build our own kind of statically compiled ruby which you can just use anywhere that's
16:34 auto installed for you and everything so i guess uh some of the something else that i'm wondering is
16:41 are you taking advantage of some of the newer features in ruby so for example you've got
16:46 yjit right that speeds things up quite a bit or uh the other one is like uh fiber scheduler and stuff
16:52 like that which gives you some level of concurrency or are you digging into any of that or is that kind
16:57 of upcoming when you get to it yeah so we've we've played around with like bits and pieces around
17:02 there like so it's it's funny because homebrew is very much you know that there's been this kind of
17:07 meme in the ruby community for a while that like people get sad that like ruby is associated almost
17:13 entirely with rails most of the time um and like homebrew is pretty much as as far away from that as you can
17:20 get you know we're not we're not even a web app we're not attached to a database like you're running
17:25 a cli so it's funny um i i guess a lot of the places from where the jets come from like it never
17:32 say never like it may well make things faster in the future but right now all of the kind of jets
17:37 different jets we've tried in the different jet sessions we've tried like just the way homebrew runs
17:42 even if it's doing a relatively complex task like a this episode is brought to you by spreaker the
17:48 platform responsible for a rapidly spreading condition known as podcast brain symptoms include
17:54 buying microphones you don't need explaining rss feeds to confused relatives and saying things like
17:59 sorry i can't talk right now i'm editing audio if this sounds familiar you're probably already a
18:05 podcaster the good news is spreaker makes the whole process simple you record your show upload it once
18:11 and spreaker distributes it everywhere people listen apple podcasts spotify and about a dozen
18:16 apps your cousin swears are the next big thing even better spreaker helps you monetize your show with
18:21 ads meaning your podcast might someday pay for well more microphones start your show today at
18:28 spreaker.com spreaker because if you're going to talk to yourself for an hour you might as well publish it
18:35 the run times are relatively short the process only sticks around right for you know at most probably
18:40 a few minutes um but then a lot of the time it's just around for a second and there's new data every
18:47 time so you can't necessarily cache things super hard um and like basically everywhere where homebrew is
18:55 slow and like that's the most common complaint about homebrew is that homebrew is slow um the almost
19:00 everywhere it's slow it's slow because of um the like io basically like we're either downloading
19:07 something off the internet we're writing something to your your hard drive we're reading something from
19:10 your hard drive whatever and we can maybe do a little bit more parallelization but again like in the past
19:15 uh there's been times where uh there's been times where i've and again some of this may be changing
19:19 from spinning platters to ssds or whatever there's been times where i've ripped out paralyzed code that
19:25 we have in homebrew and it's actually dramatically faster running really than it is parallel um so yeah
19:32 i mean we we we definitely have a decent amount of like profiling tools like we rely on like stack but
19:37 basically everything you would expect in rubyland stack prof co-prof all these types of things
19:41 um but yeah but we do try and make things as fast as we can and the jets don't seem to really be
19:47 helping much for us at the moment but as i say maybe yeah so um i'm going to move on a little bit
19:57 to you mentioned homebrew core so is that the list of casks and formulae so homebrew core is the list of
20:05 formula so we have homebrew cask and homebrew core so the difference between casks and
20:11 formula are like homebrew costs used to be a certain like separate project and that's the way
20:16 that's what you use for installing like binaries from other people so like right like google chrome
20:22 or one password or whatever like if you're installing that basically if if the project is
20:29 whether it's open source or not if if the main way that that's delivered is like a pre-compiled
20:34 thing or like a dot app or dmg or dot pkg or whatever then you're installing that with homebrew
20:39 cask like the essentially to the user the interface is more or less the same nowadays like you type
20:44 brew install google chrome or brew install wget and it's it looks essentially transparent under the
20:49 hoods but right yeah they're they're managed in separate repos like this was previously like a
20:54 requirement due to some open source political nonsense um but yeah so everything in homebrew core
21:00 is open source software and everything and homebrew cask has a much looser uh requirement basically there
21:07 um so yeah they're managed separately they have roughly the same amount of packages but then like
21:11 again the the dsl that's used are kind of slightly different and homebrew cask is a bit more friendly
21:16 to like proprietary software for guis whatever and then homebrew core is a bit more friendly to like
21:20 clis and there's open source software i gotcha so yeah and see i didn't even know you could install
21:29 some of those binaries with homebrew because i always just go to the website and download it and
21:34 run the installer and right yeah or open the dmg and drag it to applications or whatever so that that's
21:42 really interesting um as you get into some of the open source stuff in particular how do you know how to
21:49 build it are you always looking for the make file or is there something else to it or so sometimes there's
21:54 instructions in like a github repo or whatever or sometimes there's kind of you know it's got to
22:00 make file and sometimes you just kind of try it out and figure it out and hopefully you can do the
22:07 right thing and then two years later the the author pops up and they say hey you're building this totally
22:11 wrong and it's like well you didn't tell us what to do dude so like that's that's what's gonna happen
22:17 yeah so yeah it's it's always a like a there's definitely like uh a certain amount of kind of
22:24 trends and traits of things that behave pretty similarly and those projects are very nice to
22:31 deal with and then there's projects who decide hey i'm gonna invent my own build system because
22:37 all the other ones suck and yeah those are less fun to deal with right so um essentially if i do a brew
22:46 install on something like google chrome it's just gonna pull the binary off of wherever and it's
22:56 gonna put it wherever it goes and make sure it's in my path or whatever whatever it has to do in order
23:01 for me to be able to run it yeah exactly and then for a formula it's going to go through some kind of
23:07 build process and it sounds like you don't like auto detect uh the build script you go and you look it
23:15 up and you add that to the formula and tell it how to build it yeah exactly so what we we try and do
23:21 like certain bits of water detection and for an end user for the you know if you're running brew and
23:25 saw on your machine uh the experience is actually kind of similar because you're you're still downloading a
23:29 binary most of the time if you installed homebrew on like a supported macOS version which is one of the
23:35 the last three and the kind of default location right then what it will do is essentially when
23:41 someone submits a formula or an update to a formula we have our pull requests on github
23:47 our continuous integration we'll go and like build that and then when that works we then
23:50 have a process that kind of that build artifact gets uploaded we call those like bottles the binary
23:57 packages and okay like that that was my probably favorite contribution to homebrew was doing that
24:04 just for the amount of time it's probably saved the world uh because before that you know if you
24:09 wanted to install wget it would take five minutes and everyone would essentially spend five minutes with
24:14 their cpu fans you know in the days before right silicon like going 100 to spit out essentially the
24:20 the same binary at the end um so yeah now we kind of have these binary packages and those are downloaded
24:27 on your machine and installed and it's all nice and and also it's way less error prone because um with
24:33 the compilation stuff there's always things where you know you have some random utility on your machine
24:38 and the build script detects that and then blows up or whatever so you don't have those same issues with
24:43 the binaries and hey there this is charles maxwood i'm excited because i wanted to let you know about
24:51 this thing that i pulled together that i had just i've been dying to have this for years and i never
24:57 felt like i could and then i just realized that there's no reason why i can't so um i'm putting
25:01 together a book club and we're going to read development focused books career books you know
25:06 technical books whatever the first book that we're going to do is going to be clean architecture by
25:12 uncle bob martin if you're not familiar with clean code or some of the other stuff that bob has done
25:17 check that out i've also talked to him on the clean coders podcast which is on top end devs
25:21 but uh yeah we're going to get on he's going to show up to some of our meetings and what i'm
25:25 thinking is we'll probably have like five or six people uh part of the conversation along with bob and
25:30 i at the same time and we'll just uh so somebody can come on they can ask their question
25:36 and then we'll just rotate people through so we'll we'll mute one person unmute another person
25:41 when it's their turn to come on and be part of the discussion so we'll do that for like an hour
25:46 hour and a half and then the other part of it that i'm putting together is just kind of a meet and
25:50 greet gather area on gather town and so after the the meetup and the call what we'll do is we'll all
25:58 go over to gather town and you can just log in walk up to a group and have a conversation and that
26:03 way we can all kind of get to know each other and and make friends and and get to know people
26:08 across the world uh one thing that i'm finding is that yeah the meetups are starting to come back
26:12 but a lot of people don't have the opportunity to go to a meetup and i really want to meet you guys
26:16 and talk to you so we're going to put all that together it'll all be part of that book club you
26:20 can go to topendevs.com book club to be part of it and i'm looking forward to seeing you there the
26:25 first book club meeting will be in december the beginning of december we're starting the first week of
26:30 december and um you'll also be part of the conversation about which book we do next i have
26:37 one in mind but i want to see where everybody's at so there you go right so it looks at my setup and
26:44 says oh i know that this bottle's compatible so i'm just going to give you the binary instead of
26:50 making you build it totally and then if it can't for whatever reason then it'll do the build step
26:56 or the build process yeah yeah and i mean that that essentially should never happen on you know
27:03 a supported mac and a supported location but yeah if you're outside of that then it will fall back to
27:08 kind of building from source and when it can't do that otherwise uh but yeah but most most things
27:14 most of the time will always be you get like a binary package for most users so right so just backing
27:23 up a little bit because you mentioned homebrew core and homebrew casks is kind of the places where
27:29 these things live is homebrew literally going to github and cloning it or like how does it get that
27:36 list yeah originally well until very recently yes we we did that uh to get hubs yeah so the original
27:45 updater for homebrew was just it had one repo uh which was like mxcl which is max house github uh
27:53 mxcl slash homebrew and that would just be cloned in a directory and when you ran brew update
27:58 to like you know pull the latest packages that was just doing a git pull basically um so over over
28:03 time that became more and more sophisticated um as homebrew grew that became slower and slower like
28:10 even the no-op operation of just doing a git fetch on that repo and it's telling you you don't need
28:15 anything that was taking like a couple of seconds at the time i worked at github and i was getting
28:20 annoyed with how slow it was so i actually added a entry to the github api to specifically to make
28:27 homebrews updates faster and uh like it ended up being used by a bunch of other cases and package
28:33 managers or whatever essentially what you could do is you could just like pass the the hash that you
28:37 have i i have used the the uh what's called the http e tag so you could pass the the git commit hash as
28:46 the e tag and then you would get a whatever i can call it is the the status code that says no not
28:52 modified essentially so it would do that and it also wouldn't count against your github rate limit
28:57 because that was very heavily cached that operation so that basically meant you could get a really really
29:01 fast response for like github repos and we still rely on that api to this day um and yeah so until
29:08 very recently like that was still how it was done and we would like fetch stuff or whatever but um
29:13 over time that was getting slower and slower and also like github were like you know they were pretty
29:18 nice about it all but i think at some point i believe there was like a dedicated bare metal server
29:23 that was essentially just 24 7 serving homebrew core to people uh because also the way that that
29:29 that repo had super long history loads of contributors right didn't i did a lot of like uh squash and
29:36 rebasing so like not very kind like history management um and again like thousands of files in a folder
29:44 which we never knew when humbrew was created is extreme if you have very deep histories get gets very
29:50 angry and the performance goes to um bad places pretty quick this episode is brought to you by
29:56 spreaker the platform responsible for a rapidly spreading condition known as podcast brain symptoms
30:02 include buying microphones you don't need explaining rss feeds to confused relatives and saying things like
30:08 sorry i can't talk right now i'm editing audio if this sounds familiar you're probably already a podcaster
30:15 the good news is spreaker makes the whole process simple you record your show upload it once
30:20 and spreaker distributes it everywhere people listen apple podcasts spotify and about a dozen apps your
30:25 cousin swears are the next big thing even better spreaker helps you monetize your show with ads
30:31 meaning your podcast might someday pay for well more microphones start your show today at spreaker.com
30:38 spreaker because if you're going to talk to yourself for an hour you might as well publish it
30:44 um so yeah so as of i think it was year and a half maybe two years ago uh like someone one of the home
30:52 room maintainers kind of started the first attempt at like a little project to see about okay well we
30:59 can we you can effectively turn pretty much all the homebrew packages into like json output anyway right and
31:05 when you're just downloading binary packages you're basically just downloading a thing off the internet
31:09 and sticking it in um like installing it that way anyway you don't need all the kind of sophisticated
31:15 build information for every package um so yeah so as of yeah like about a year ago i think we've
31:21 migrated entirely to what we call our kind of json api which is downloading just a single json file which
31:27 provides all the details for those packages um and then you just have that single download you don't
31:32 need to have that git repository on your machine anymore which saves a bunch of space and it's like a lot quicker
31:37 of downloading um right and yeah and the implementation of that api is another disgusting uh thing where so
31:44 perhaps because i'm scottish um we are known for such things or just i don't know uh i don't like
31:50 paying for things and and in open source projects i think that's universal even less uh so our our json api
31:58 is actually on github pages um and it's yeah so it's all like it's a statically generated api essentially
32:05 so like every 15 minutes we like regenerate our packages and what that means is that it's really
32:10 really really fast and essentially you know nigh infinitely scalable for github to stick a cdn
32:16 in front of that and host it that way and and it's also pretty easy for us to kind of download and
32:21 debug and we're not maintaining some like complex um like you know when i worked on the the api github
32:28 like maintaining the performance on that thing when you were at the scales we were at was not an easy
32:36 feat and yeah and i i saw that i guess for homebrew and i was like nope i'm gonna if i can make this
32:41 just entirely static unauthenticated api then let's do that so that's the route we went down and it's been
32:47 working pretty well for us that's awesome i love that it also makes me think a little bit that
32:55 sometimes i might be over complicating my life by trying to make it work the other way
32:59 right the more traditional i'm gonna build i'm gonna build an endpoint on rails and it's gonna
33:04 give all this stuff and it's gonna check all these things instead of just saying hey it doesn't matter
33:09 if everybody knows this and it'll be way faster and it's not that big a file so yeah just
33:17 go get it yeah it's funny because i had this idea for like over a decade about like building
33:22 essentially static apis where because you know like static site i've used like jackal and github
33:29 pages for my oh yeah for a long time um i i kind of like that sort of static site builder model and
33:35 you know the results are nice and super fast and it's really cheap and easy to host and very
33:40 non-error prone right once once you've generated the files you're like well nothing can go wrong now
33:45 right compared to like maintaining a rails app um but yeah so i always kind of like the idea of having
33:51 something behind i was i was pleased that i was actually finally able to do it because in cases
33:54 where essentially you want to deliver the data to absolutely everyone the data is the same it's not
34:00 like user dependent whatever and the data only needs updated like not very often like you're not updating
34:06 like second by second from some database but you're updating it like you know a few times an hour then
34:11 yeah like this approach has actually been really quite cool and fun and yeah and encourage others
34:17 to play out below uh play around with the approach and maybe not on get on pages because get on my gang
34:22 room so well i can imagine though because i'm just thinking through maybe what i want to do with
34:29 building an app for the podcast network right yeah and yeah like 99 of the stuff that i'm putting out
34:37 there it's going to be the same for everybody and it's all public and right so i i don't have to protect
34:44 it at all and so yeah why not just have a static file out there that it can just go oh you want the
34:50 list of podcasts here right oh you want this podcast and all of its episodes here right now granted some of
34:57 the podcasts have more than 500 episodes at this point and so you might get a little longer json file
35:03 but still it's not terribly large and you know you start getting into it really when you start pulling
35:09 down images or audio files yeah you know i can put those in as urls on the json and just make it really
35:15 wicked fast and then it's okay now i have restricted access for memberships and stuff like that that's where i can
35:23 you know maybe put a check in front of it but still then just have them pick up a statically generated
35:29 thing because once you're authenticated that's all the same too so yep totally i'm really digging this
35:35 yeah it's it's been fun i think i don't know that this is the thing i like about working with open
35:40 source and stuff like homebrew is that you know you get i guess the thing i've learned in my career is the
35:46 most interesting projects are the ones often with any sort of constraints right like the projects would
35:52 like you know take as long as you want you have as many people as you want do it whatever you want
35:58 in whatever language you want like they're they're often like maybe i'm just a masochist but i i find
36:02 them often like less fun whereas the fun thing with something like homebrew as well it's like you're
36:06 building homebrew's api and it's like well we we have like i i guess the way i describe homebrew's
36:12 financial situation is like not enough to pay anyone but too much for stickers right so you have
36:17 like this like middle ground of money uh and as a result like it's this tricky thing of like well i
36:25 don't really want to put us in a situation where we're paying vast hosting bills so let's try and
36:30 figure out a way of doing this for free essentially but it has to be able to scale to like you know
36:36 millions if not tens of millions of people like using this thing um so yeah so like that's been a
36:42 a fun kind of comparison uh compared to as i say you know like github days where you're scaling to
36:48 many more people than that but you also have particularly post microsoft acquisition like
36:53 many more resources to your disposal in order to achieve those scales and you're not working quite as
36:59 much as much about the bottom line as you go along yeah yeah the other thing that i i just went through
37:06 my head was and this isn't json it's xml but rss feeds right oh yeah yeah yeah yep so yeah um
37:15 all right so we've kind of talked through how the repository works and um let's say that i decided i
37:24 wanted to build the cli and it was moderately complex are there are there things that you
37:29 recommend to me as a ruby developer as places to get started um i mean the funny thing is i i wasn't
37:39 you know i realized this uh maybe um i don't know a controversial take for the podcast but like i mean
37:45 nowadays i probably wouldn't necessarily reach for ruby for a cli like it's it's okay for like doing
37:52 doing you know i still think it's great for doing web apps but like you know when i've i've been
37:57 building like a new cli for some of the stuff we've been doing with work brew um and like go feels like
38:04 a more i was basically weighing up between like go rust and swift basically um as oh okay like
38:10 the switch i wouldn't have pegged swift for that well i yeah i i didn't peg swift in the end for
38:15 various reasons but you know i ended up going with go which i it's i don't know like
38:21 for me go is just it's good at what it does it's such a boring language i i don't have any fun when
38:28 i'm writing go but you can ship a single static oh it's fast yeah and it's super fast and for cli
38:34 it's like it's like the responsiveness is really great i think i think that's the thing that i maybe
38:39 i'm just burned for like 15 years of working on a ruby cli but like there's a certain amount of
38:43 people will complain about the slowness that it's like literally even like a ruby interpreter spinning up and
38:49 saying hello world is like unacceptably slow to some people right and at that level i can't
38:55 really do anything to make it it faster but yeah i guess if you if you are going down the route of
39:01 almost like really wanting to get into a make it some sort of ruby cli or whatever i guess i would
39:05 just say like you know the beauty of ruby is that you can start with like a single script right you can
39:11 just right have a script stick the shebang at the top like the for those who don't know that's the thing
39:16 where it has um like exclamation mark hash like user bin anyway like user bin ruby or whatever and then
39:26 you can just start writing ruby code straight away right and and like that's that's how i think is the
39:33 best way to maybe get started with like clis whatever it's just like rather than almost like
39:37 reaching for some option parsing framework whatever no just go and just start writing some ruby and right
39:44 just read the argv which is the array of like the arguments you pass into the script and like just
39:49 read that yourself and just you know don't don't go too fast too quickly and you'll probably have
39:55 like a good time because again like when you're clis if you're wanting to distribute them to other
40:00 people like if they you know if you start relying on loads of other gems and stuff then it becomes like
40:06 a little bit more of a headache to kind of distribute and use these things and keep them up to date and
40:10 whatever so like yeah i think the beauty of ruby is like ruby has like a pretty decent broad standard
40:16 library you could do a lot with like a single script and even like an old version of ruby like the ones
40:20 that ship with like mac os like there's still you know it's still a powerful like pleasant language to
40:26 work in so yeah very cool um so yeah so is there anything new coming in homebrew or
40:36 yeah i mean we're always working on this it's funny because uh people quite people ask periodically
40:43 about like the roadmap for homebrew and because we're essentially i mean maybe except for me now
40:48 like essentially almost everyone is like a volunteer the the roadmap for homebrew is kind of like
40:52 well let's see what people want to work whatever people want to work on right and it's yeah like
40:58 the product management on homebrew kind of looks more like uh nerds attempting to nerd snipe other
41:04 maintainers into um solving problems than it does like you know building a team and assigning them a
41:10 six-month project right so yeah i mean i think like a big thing is like performance like we're continually
41:16 trying to kind of eke out better performance where we can um and then yeah and i guess like i'm looking
41:21 at it from the kind of work brew end as well of being like okay well so if uh if me and a team of
41:28 people are actually able to work on homebrew full-time or like spend like a decent amount of time putting
41:34 into homebrew like what are the really big impactful things in homebrew that kind of might people might
41:40 look at and say that's just impossible that we could do if we have like a paid team of people who are
41:45 working on this stuff full-time um and i don't want to you know i don't want to be too spoilery with
41:50 that but um yeah there's definitely there's definitely some stuff we're exploring where
41:54 if we see that there's interest and we can kind of pull it off i think people will really say wow about
42:01 but um i in the short term medium term i think there's more along the lines of yeah just like
42:06 making homebrew faster more secure making the contribution process better and just trying to
42:10 do everything we do right now like better basically like i mean i i'm a big fan at github and on homebrew
42:17 of like um when i was at github that was like reading really really heavily into automation so again
42:23 like a lot of what we're trying to do on homebrew sometimes it's like we've only got 30 people
42:28 maintaining the the project but yeah the throughput of like prs we have and it's fairly ridiculous and we
42:34 just try and have so much automation lean have more and more and more heavily into that and any time
42:39 you know i catch people doing things that i'm like a bot could do this then it's like right i try and
42:44 write some code and get the humans doing the human things and save the the bot work for the bots
42:51 so i was gonna ask about work brew and i'm gonna ask about in a second but you did mention um
42:57 uh maintainers and so if somebody is like oh i've got a great idea for homebrew or um you know i use
43:07 this all the time and i'd like to contribute back to it right is is there a good way for people to get
43:12 started with contributing yeah so homebrew the best way to get started is there's like a contribution
43:17 section on the readme of the homebrew brew so if you go there then you'll or if you even just search
43:23 for like contributing to homebrew like the result like results are pretty decent i would say our our
43:27 docs are probably like on the better end uh for like contribution like our tooling is probably pretty
43:33 good but like i would say like it's a really easy project to get started with um and we we really
43:40 try to kind of cheerlead the people who want to get involved with the project so like if you're in
43:45 any doubt just try something and we'll help you out and you can mention me by my github handle or
43:51 whatever if you say that it's your first pr and you're struggling and i'll try and do what i can to
43:54 help you and stuff like that um it's a yeah i as i say i think it's a really good project to kind of
43:59 get started with um but as you know for some of us at least it's also can be uh fairly addictive
44:06 at times so be be wary from that perspective as well yep all right so the other question i have
44:14 then is yeah do you want to talk about work brew like what it is and where you're heading with that
44:20 yeah so i mean as i mentioned earlier with like homebrew like i've been working on homebrew for
44:25 about 15 years and for you know probably 10 of those years there's been people who have been
44:31 encouraging me at various points that oh you know you should make a company doing stuff around homebrew
44:35 and i i have been historically kind of down on this idea or fairly cynical particularly when
44:42 i mean i'm not going to name any names but you know there's been various companies who kind of
44:46 have done open source stuff and then you know they complain down the line when everyone uses their
44:52 software but they're not making enough money or whatever it may be and right i guess to me it's like
44:57 open source itself is not a business model i know and also like i you know i kept like you mentioned
45:03 earlier like you've got you know kids younger than homebrew and stuff like i i also feel like homebrew is
45:08 kind of my first child and i feel quite precious of it and i wouldn't want to you know like destroy it
45:14 or whatever so it's it yeah it took a while for me to get to a place that i was like yeah actually it
45:19 feels like a few things have come together so one is like homebrew feels i mean even with the road map
45:25 stuff like if we'd had this conversation a year ago or two years ago i probably would have said like
45:29 more stuff on the horizon but it feels like fairly close to kind of feature complete at this point
45:34 and without almost like a big um injection of kind of effort and stuff so that was one part the other
45:40 part is like we've got i won't go into all the boring details of like open source governance but you
45:45 know but we have an open collective we have a like a leadership committee we have elections every year
45:50 we have like ways of getting involved with the project beyond just like current docs and stuff
45:54 like that we have money that we're using to like provide stipends to some maintainers and pay for
46:00 them to kind of come and meet each other in person and stuff so the project feels like it's in a really
46:04 good mature place and it's also in a place where no one party could ever kind of completely take
46:10 over homebrew even if they wanted to and so yeah so and it it's also on the backdrop of this i guess
46:16 like you know i left github last year and was sort of looking for a new challenge or something new and
46:22 you know again over the last 10 years there's been various times where companies come and say hey
46:26 we'd like you know homebrew to do this or that or whatever and this episode is brought to you by
46:31 spreaker the platform responsible for a rapidly spreading condition known as podcast brain symptoms
46:37 include buying microphones you don't need explaining rss feeds to confused relatives and saying things
46:42 like sorry i can't talk right now i'm editing audio if this sounds familiar you're probably already a
46:49 podcaster the good news is spreaker makes the whole process simple you record your show upload it once
46:55 and spreaker distributes it everywhere people listen apple podcasts spotify and about a dozen apps your
47:00 cousin swears are the next big thing even better spreaker helps you monetize your show with ads
47:06 meaning your podcast might someday pay for well more microphones start your show today
47:12 at spreaker.com spreaker because if you're going to talk to yourself for an hour you might as well
47:18 publish it essentially what a lot of homebrew features come down to is like do the volunteers
47:22 who are mainly working in their spare time want to build and support this right and for a lot of like
47:28 big enterprise stuff i don't know charles have you ever had the misfortune of uh reading through like
47:33 a saml spec or something like that before but you know once once you get into this like big enterprise
47:38 tech like it's often not very fun and it's not what people want to spend their evenings and weekends
47:42 doing and i previously worked like before my homebrew days in like an open source company back when i
47:47 was a kde contributor and this company uh consulting company and they hired more kde contributors than
47:53 anyone else and i worked and i thought sweet i'm gonna be able to work on open source most of the time but i
47:57 i learned pretty quickly that their open source projects were actually not super fun because they
48:01 were basically stuff that people would pay a consulting company to go and like hey like we've got this
48:06 stuff which we can't get anyone to do for fun so will you do this for money and then they're like okay
48:12 well fine um so with work brew i kind of see it as being a similar thing where what we're trying to do
48:18 is you know there's a lot of talk about sustainability and open source and stuff right now um and i guess that's
48:24 on my mind i worked on the original team of people who built get up sponsors and have helped kind of
48:29 homebrew yeah have that open collective and to me like this is the next almost like step in the
48:33 evolution of like what sustainability could look like for homebrew is having a commercial entity where
48:39 we're trying to kind of make a sustainable like commercial entity who has an interest in making money
48:45 in the homebrew ecosystem but not charging money directly for homebrew itself homebrew itself will
48:50 is and will always remain like free of charge but we're going to make homebrew better by making
48:55 workbrew better and be providing services to companies um in the homebrew ecosystem through
49:00 our expertise and by building stuff that tightly integrates with homebrew so essentially the version
49:05 of workbrew that i like run on my local machine today it looks and behaves almost exactly the same as
49:10 homebrew right except it's got a little coffee cup instead of a beer emoji um but it's you know it's
49:16 running in a different way it's running in a more isolated like fashion a multi-user fashion right
49:20 it's like reporting back to do fleet management so someone could upgrade like if i've got critical
49:26 vulnerabilities remotely and so that people can see kind of oh like you've um you haven't updated in six
49:32 months and we've got some problems with that whatever it may be um so yeah so essentially that's what we've
49:37 tried to do we're trying to build something that provides a better experience for using homebrew at work
49:42 um but for most hobbyists i i mean we very much have the goal that if you in five or ten years if
49:48 we're successful even if you've never heard of workbrew even if you never interact with workbrew you will
49:52 look at essentially roundabout now as being an inflection point where homebrew gets really really
49:59 good in the next five or ten years and it gets noticeably better for everyone using it because we're
50:03 able to kind of get involved and improve the sustainability of the project through the commercialization of this
50:09 part of the ecosystem that's cool um if people want to know a little bit more about
50:17 what's going on with that like you have a mailing list or something or yeah so date somehow yeah the
50:24 best way of like getting in touch with us we've got a contact form on our website if you go to workbrew.com
50:29 contact and then you can like send us a message and then we can go and get in touch and we'll kind of
50:36 send you emails and stuff like that and you can also see if you go to workbrew.com our demo of
50:40 what we have in private beta right now which is essentially like a fleet management for homebrew
50:45 and kind of integrated with mdm tools and so that basically like mac admins and primarily and you
50:51 know security professional developer experience folks can basically get a nicer experience of using
50:55 um homebrew at work and so yeah so we're recruiting like design partners for that right now which
51:01 essentially means like people who will give us lots of feedback while we make the beta kind of
51:06 more to their requirements so again if you're interested in that best way is reaching out through
51:10 the contact form and and if you have any thoughts equally as well if it's easier if you're one of
51:15 these people who gets spooked by contact forms you can send me an email um my email address is pretty
51:20 easily findable on the internet so if you do that or it's just mike at workbrew.com or at brew.sh if you want my
51:27 open source email awesome all right well um if people want to connect with you on some other level
51:36 where do they find you on the internet so i'm still i'm basically read only mainly on twitter nowadays uh
51:45 i'm mike mcquade on there but the best place probably is uh i'm mike mcquade at mastodon.social
51:51 on like mastodon nowadays um or if you just find my website at mike mcquade.com that's basically got
51:56 you know my i try and kind of put if i write blog posts or like do talks and all my contact
52:04 information is there and stuff as well so yeah and you'd be surprised how few people actually just
52:10 email me to say hi or share their opinions so feel free to do that for almost anything except for
52:16 saying my home is broken please fix it in which case right all right good deal well i'm gonna go ahead
52:23 and uh roll us into the picks here have you ever wish that you had a group of people that were just
52:28 as passionate about writing code as you are i know i did i did that for most of my career i'd go to the
52:33 meetups i try and create other opportunities and it was just really hard right the meetups i got some
52:39 of that but they were only like once or twice a month and it was just really hard to find that group
52:43 of people that i connected with and and really wanted to you know talk about code a lot right
52:49 i mean i love writing code i think it's the best and so i've decided to create this community and
52:55 create it a worldwide community that we can all jump in and do it so we're going to have two workshops
53:01 every week one of those or two of those every month there are going to be q a calls right where
53:06 you can get on you can ask me or me and another expert questions uh the rest of them are going to
53:10 be focused on different aspects of career or programming or things like that right so it'll go
53:16 anywhere from like deployments and containers all the way up to managing your 401k and negotiating your
53:22 benefits package well we'll cover all of it okay and then we're also going to have meetups every month
53:28 for your particular technology area so we have shows about javascript react angular view and so on
53:34 we're going to have meetups for all of those things i'm going to revive the freelancer show
53:38 we'll have one about that right so you can get started freelancing or continue freelancing if
53:42 that's where you're at and i'm working on finding authors who can actually do weekly video tutorials
53:49 on something for 10 minutes this related again to those technology areas so that you can stay current
53:54 keep growing so if you're interested go to topendevs.com/signup and you can get in right now for 39
54:03 when we're done that price is going to go up to 75 and the 39 price gets you access to two calls per week
54:12 the the full price at 150 which is going to be 75 over the next few weeks that price is going to get
54:21 you access to all of the calls and all of the tutorials and everything else that we put out from
54:26 top end devs along with member pricing for our remote conferences that are coming up next year so go check it
54:32 out top end devs.com/signup um and i we usually have the host go first so i'll go first um one of the
54:42 first picks i have i always pick a board game when i uh when i do my picks and so i'm going to pick a
54:48 board game called fire tower and basically it's a it was a relatively simple game let me look it up on
54:57 board game geek real quick while i tell you how how it's played but um essentially all of the players
55:03 have uh fire towers in the corners of the board and
55:07 uh so what you're trying to do is you're trying to burn everybody else's tower down before the fire
55:16 gets to and burns your tower down and so um you can play two to four players we played it with four
55:24 people it was a lot of fun uh board game geek rates it at 1.77 uh i tell people that a two is
55:30 kind of a reasonably complicated game it's complicated to be fun but not so complicated that it's hard to
55:36 pick up and so yeah this is right in there at that level um playing time it says 15 to 30 minutes i think
55:43 we played at 45 but it was our first time um uh yeah 10 plus player or 10 plus on age i honestly think
55:53 my eight-year-old could play this game right um it's it's pretty simple and after she played it once
55:58 or twice she'd probably get most of the strategies figured out right don't have the wind blow toward
56:03 your tower um put put down the fire tokens in a way that gets it closer to the other people's towers
56:11 i mean right anyway so um so i'm gonna go ahead and pick that uh it was a super fun game really
56:19 enjoyed it um one other one that i'm gonna pick is we were talking about some of the install stuff
56:27 and daniel kehoe who's used to be in the ruby community way back in the day uh he has a website
56:34 mac.install.guide and it walks through how to install like ruby and a bunch of other stuff and it
56:40 pretty heavily uh refers people to homebrew so um anyway i'm gonna refer to that just as a resource
56:49 that people can uh go check out and then um
56:54 i had something else and i'm i'm just kind of blanking on it anyway i'm the the the where i talked
57:03 to daniel most recently was when i picked up the rails composer.com domain from him
57:07 um and so um i'm working on that you can actually watch me build rails composer and all of the
57:14 different uh rails engines that i plan on putting together for that um on the rails clips um
57:24 video series so just go to topendevs.com you can find it i think i also own rails clips.com and i think
57:31 that goes to the right place so uh go check that out but uh anyway um yeah good stuff uh mike what do
57:40 you want to pick on the show yeah okay i guess i will do three little picks uh so one would be i'm
57:48 reading a book right now by a guy called alistair reynolds and he's a science fiction author he's
57:53 written a lot of kind of space opera stuff i've written read all of his books uh that he's ever
57:58 written so far um and he released a new one a couple of months ago which i'm 88 of the way through
58:03 called machine vendetta that you penny arcade talked about it um and you probably don't want to jump
58:08 straight into uh machine vendetta as there's been a couple of books preceding it in the series
58:13 unless you want to deal with the discussion of hyper pigs in the first paragraph which may throw you
58:18 off um my second recommendation would be a band uh you may or may not be into their vibe they're called
58:25 ale storm uh like a storm of beer and i guess you can see there's a theme going on they are a scottish
58:32 pirate metal band um i'm seeing them live uh next sunday uh in scotland uh but they play internationally
58:40 and they my kids really like a song from their latest album called uh pirate party because it has a
58:47 a moment with a man drinking from a shoe which is apparently very funny when you're four and six
58:51 years old uh but generally i would not recommend most of their songs to children and as they are
58:56 very heavily expletive written in some of them but you but you know whatever they may work for you or
59:03 not and my final recommendation i guess on the theme of um open source maintenance in particular is
59:09 therapy i've been doing therapy intermittently since 2020 and i've something i found tremendously
59:16 helpful personally um with all sorts of things in work personal life my uh relationships with my
59:24 friends family kids wife and and even on occasion i have sessions where i talk about a particular
59:31 difficult person i'm dealing with open source land uh in a given therapy session so if you're if you're
59:36 interested i would highly recommend it and i've written a blog post on my website of like a step-by-step
59:41 guide if that's your way of doing things of like how to find a therapist so if you are interested and
59:46 that's useful to you please consider doing that oh man given the last few months that i've had
59:54 i might need that anyway uh what was the name of the band again and if you can put a link into the
59:59 comments that'd be great yeah yeah i will drop a link into the band obviously um but yeah we'll go ahead
1:00:07 and wrap up here um one other thing i was just going to put out there this is the thing i forgot
1:00:11 is that uh my contract work has slowed down a bit so if you're looking i'm either looking for a full-time
1:00:18 job or a contract prefer contract but you know if you've got an interesting set of problems and
1:00:24 you're gonna pay me well to fix them then i am happy to consider full-time work um you can just
1:00:29 email me chuck at top endevs.com all right thanks for coming mike this was really really interesting
1:00:37 thanks for having me chuck and have a good rest of your day everyone yeah you too max out everybody