Latest / Talk Commerce / Mark Shust on Building a Modern PHP Framework

Transcript

Mark Shust on Building a Modern PHP Framework · Talk Commerce

• Brent Peterson (00:01.07) All right, welcome. Today I have Mark Shust. He is the founder of M.Academy and so many other things, Mark. Just give us your 10,000 foot view of how you got started, what you're Give us a tip, Mark, about the Academy and all that fun stuff. Mark Shust (00:14.765) Sure. Mark Shust (00:18.999) Yeah, I've been in Magento since the early days of 1.4, I think, right? I caught wind of it and got recruited on, got kind of trained by the Magento team at the time. It was Magento Inc. And yeah, just picked up that super complex framework. Been involved with it ever since. Magento 2 came out, I had a hard time. Getting a setting up a development instance of it. So led me down the path of Docker Magento, which I've built and maintained now for like seven or eight years and Along that way I realized that other people probably have trouble Learning the framework. I had a really tough time with Magento 2 even though I was really experienced with Magento 1 and Yeah, I thought others probably have the same same issues so A years later, I started M Academy and it's a place that has self-paced courses, a developer community, some tools built around Magento. And yeah, I have a all-inclusive university package where you can enroll. It's like an all-inclusive membership and an AI tutor. And I've been doing that for last six or seven years. Brent Peterson (01:39.648) Yeah, that's awesome. obviously, well, you're very well known in the community. And you have helped with thousands of people, I would imagine, just kind of getting through Magento and learning how to do it, become better developers. Talk a little bit about kind of your vision and goal of Academy and Mark Shust (02:03.757) Yeah, yeah, it's meant to just kind of take that complex framework and drastically simplify it. So really anyone that's kind of a somewhat experienced programmer can pick it up. I had trouble with learning it completely. There was no documentation or it was documented, but it was kind of fragmented and very confusing and hard to follow along. And Adobe has training out there, but it's thousands of dollars. Very expensive and it prices out like small on medium-sized businesses that want to onboard and train their their Magento developers So yeah, I noticed that gap and I create Completely self-paced courses lessons within it are very streamlined and simplified They're all usually between one and five minutes with some exceptions for some courses but each lesson kind of stands alone and teaches one specific topic and it's done in a what I call linear learning approach where you it doesn't introduce a new topic or multiple new topics at once without you knowing all of the prerequisites of the previous lessons. So it's built in an order that just makes you helps you understand it really quickly and simply and right away. Brent Peterson (03:23.01) That's perfect. All right, well, we're really here to talk about is, we'll call it Magento 3 or Marko. guess I'm not sure what you want to call it, you have something that is released today, April 1st, that is super exciting. So tell us about your inspiration to do this and what it means and how it's configured. Mark Shust (03:45.643) Yeah, so working with Magento, I've been working with it for, I don't even know how long, maybe 15 years. So, and I'm intricately involved with all of the pain points and frustrations of having taught it for so many years and noticed kind of recurring questions that come up and issues with trying to figure things out. And Magento is very, very powerful, but it's really just an e-commerce platform too. It's not like a more general purpose framework. It is built on top of Zen, but that was so many years ago already. And Magento also ships with like the kitchen sink where like there's all these modules, right? And it's low kind of by default out of the box. There are of course other frameworks that I love, like Laravel. I've built quite a few apps on Laravel and I was actually a Laravel developer for a year before I came back to Magento land. It's like the fast and modern way to build an app in modern days, but it really lacks the modularity that you actually get in Magento. And there's really no clean way to override or extend vendor code or other module code that you install in Laravel. So I think Magento got a lot of things right with specifically the ecosystem, like the module ecosystem, WordPress and Drupal kind of have that built around in the same way with community modules and the marketplace and sharing code among many projects. And Magento also has awesome dependency injection that's, think, unrivaled compared to other platforms. And it has more modularity than any other system. But again, it's just built for e-commerce. But yeah, Magento, I think, got a lot of stuff wrong, too. It's brutally complex. No one can kind of argue against that. Sure you're familiar with it. It takes like months to onboard a developer, if not like years to master. It has a extremely large XML layer that's kind of disconnected from the code also. So you have to like creating a route, you have to set your routes.xml file, and then you have the route file. So it's like, you know, it's not completely unified and simple. Mark Shust (06:10.623) So, it forces you to do stuff like the Magento way, right? We all know like you have to follow the framework and the conventions and exactly how they lay things out. Otherwise things won't work. And error messages are also like really cryptic or like non-existent. So I heard that Magento 2 also before it came release, you know, it was under development for like a long time, six or seven years. And I think I heard it was originally supposed to be really simple. It was actually supposed to be Magento once simplified. Somewhere along the way it kind of like lost track. But I've always wondered like what is that alternate future? Like what would it have looked like if they actually did make it simpler? And what would Magento look like if I was designing a framework today from scratch using modern PHP 8.5. standards and you know what would I keep what would I throw away and can I have all the Magento's power too that it has like the dependency injection layer the modularity the plug-in system the ecosystem but without all the complexity that Magento has and you know what if like we can pretty much get rid of the XML layer completely and when you had errors that you can actually like get an actual document of what went wrong, why it went wrong, exactly how to fix it. And it might even give you something to copy and paste into an LLM. All of our development workflows have changed in the last year. And it doesn't make sense not to have that type integration with the AI and LLM layer. So I had the idea to build a brand new framework. Didn't know a good name for it. I always like naming things with an It's my name and it's Academy. So I just went with Marco because that's kind of kind of my name and the idea is just that it's a completely modern PHP framework that takes all the best ideas of Magento like the dependency injection layer, the modularity, the plugin system and rebuilds them without all of the complexity that Magento has. So that's that's like the premise. I can go into more detail if you want. Brent Peterson (08:36.536) Yeah, so I think I'm interested. This isn't sort of a refactor of Magento 2. This is a brand new Endeavor. Is that kind of what I'm hearing? Mark Shust (08:46.179) Yes, I know a lot of people have been teasing it as Magento 3, but I kind of see it as a Magento 3 in a way. It's not meant to be a complete replacement of Magento at all. It's definitely going to shift as a 0.1, which I don't think I've ever done before. It's always been 1.0 for me when I release on release day, but it's definitely going to start off as an experimental framework. I'd love feedback from everyone. out there looking at it and what does it get right? What does it get wrong? Where does this work? I'd like you to see how it works and how easy it is to configure modules and configuration and things just straight in PHP with 8.5 using new features like attributes, which I think are extremely powerful and very overlooked and seeing the extreme modularity that it actually uses. Brent Peterson (09:45.327) Yeah, and I think you mentioned too that you're not tied to any one database. You could use different types of databases as well now, which is actually the original vision I remember of Joav saying that Magento 2, you can attach them MS SQL or MySQL or Postgre, whatever database you want. Tell us a little bit about that. Mark Shust (10:05.493) Yeah, so I, I from from day one, I've always envisioned a hot swappable module piece, especially if you have the powerful dependency injection layer there where you rely just on interfaces. So you can have like a database interface and then whatever module you choose to install. Let's say I install the database MySQL module. All of the bindings will be set up in the background. So as long as you use that database interface, they could be hot swapped. pretty much as you enable the module. So same thing if I was using Postgres. I've been doing a lot of more Postgres work lately, after working with MySQL for 20, 25 years. It's because you can store vector embeddings in it, which is very powerful for LLMs and vectorized search. yeah, it's made for that hot swappable. So I can actually just, remove one module and install the Postgres module and not change any of the underlying code. Everything will just still work. when you build kind of from the ground up based completely on interfaces and with this modularized system, it allows you to do things like that. And it's true hot swap implementation. Even when you start working on a new Marko framework project, and you don't have a database driver installed by default, it'll actually tell you and ask you which one you want to install. And you can just copy, paste the code and get things installed and working. Brent Peterson (11:46.809) Do you envision this, so is your vision kind of side by side with Magento 2 or is this a drop in replacement eventually or give us a vision on where it's gonna go for e-commerce? Mark Shust (11:59.253) It's, Marco is more of the framework that I wish Magento was. So it's, I would not say it's a replacement for Magento. It's Magento has, the thing it has going for itself is it has every single feature under the sun for commerce. I, it would take years and years and years for Marco to do that. That said, I know all of our development, like processes have changed over the last year or two. If you want to ask an AI to build a Magento module and it's within a very built out platform, it's going to get very confused because of the multiple layers of architecture that Magento has, like XML and configuration and fallback and the different kind of languages that it uses. So the LLM just kind of chokes on context. So the idea... is that Marco will provide you a way to build something like Magento. So you can almost think of it like the Zen framework of Magento. It's the behind the scenes framework, kind of like Symphony or something like that, and Laravel. But you can build an app on top of it. You can build a Marco Commerce app and have all of that add to cart functionality check out and everything. And you could build it very fast using LLMs too, based on the standards and structure that are in place in the framework. Brent Peterson (13:33.102) So do you think there's a solution here? Because MySQL doesn't have vectorized search. I mean, vectorized embeddings right now, is that correct? And this could be something that sits alongside somebody's Magento 2 installation. They could then cross-reference that vectorized part of it and start using them together. some of the vision you have on this? Mark Shust (13:56.483) You definitely could. So if you are familiar with running something like Magento and WordPress side by side, I know that's a common set up. You can definitely do something like that with Marco and even gradually build something out completely custom. I think custom apps are taking off right now because of LLMs. can just ask the LLM and it'll build anything you want. So you can almost start from scratch. Here's a super, nothing's implemented. Here's a clean slate that you can start with. But here's an organized way to actually build something out that has this module ecosystem in place. So again, would probably, even for some new apps, I'd still use Laravel a lot too, because it's so awesome and powerful. It depends what my needs are. But if I want to build something that's more modular, that has a module ecosystem, like Magento, like Drupal, like WordPress, that... opens up the ability for you to share modules, it could become very powerful. And not to say that you can't build something that you currently build in Laravel and Marco, you absolutely can. And that's kind of the plan, that it's kind of a universal framework for doing all of these different things and doing it in a very simple and elegant and clean way. Brent Peterson (15:16.59) You had you had sent me a Marco talk to tell us a little bit about that what it does and how developers could use that to kind of learn the learn the framework and get started on a project Mark Shust (15:29.283) Sure, yeah, launch day, definitely wanted an app that is out there. I'll probably have other ones as well. But Marco Talk is pretty much an implementation of an app that is run on the Marco framework. So this is how a specific app, like something like Magento would look like. Of course, extremely simplified down to just demonstrating the core aspects of the framework. But Marco Talk is a place where it has spaces and you can chat with using messages, kind of like a very simplified Slack. And you could do things like pin messages on the board, react with smileys, leave messages. And what's cool about it is it's all real time. So you can have many users hooked into the same framework or this app at once and logged in. And just like Slack, you can send messages and they appear in real time in everyone else's windows. So it's a real time messaging app. But what's interesting about this is it's built completely with PHP. There are no WebSockets. There is no Node.js. It works with server sent events and it actually uses the server sent event package from Marco to do this. And the way it can do this is again, another benefit of Postgres. is that it has a listen architecture in there to send messages like PubSub. And it integrates great with AMP PHP, which is what we use behind the scenes to automatically send these messages out in real time over server set events. So it's very cool. If you've ever dealt with Node.js and PHP and it's kind of... You have all these different systems and then you have systems to maintain. need maybe a Redis database, right? For PubSub, you Node.js and a whole JavaScript app that's discombobulated from your main app. So Marco, you could just build it all with PHP. It's really cool. a way to just streamline everything and make it easy to manage as well. Brent Peterson (17:44.056) What do you say to the naysayers who say PHP is dead? I'm guessing people have been saying that for the last whatever 20 years. The life of PHP, what do you tell them today that PHP is alive and kicking? Mark Shust (17:57.483) well, I'm building new frameworks and apps myself and PHP and I, my go-to has always been, in the last few years, Laravel. don't even, I don't even consider Node.js even though I like it. I've done extensive work with WebSockets in real time and socket IO and things like that and Node. But, PHP gives you, you know, a very streamlined way to connect to the database, have authentication, have a very distinguished server side language that's pretty much built for the server. Node.js is definitely more of the client side, it seems. it doesn't, I've always had pretty bad experiences trying to build an actual CRUD app in Node.js. Maybe it's because I don't have extensive JavaScript, like super experience, but PHP just makes it simple. There's no compile. going on and you know, it's just, it's very easy to build something, especially in the age of LLMs, build something very quickly that works that can stand up and iterate on and actually ship something. And I think PHP is great at, if nothing else, it's great at shipping. So there's been so many enhancements to the language as well. Again, when Magento 2 was built back a while ago, maybe a decade ago. It didn't have things that were introduced in PHP 8.5 or any of the recent versions like attributes or constructor property promotion or, you know, just things that make your whole development experience so much better. So PHP has definitely continued to evolve and I believe it'll keep continuing to evolve. I see no reason not to think that and it's definitely great. That's my go-to language for everything. Brent Peterson (19:51.983) Looking back, think Magento 2 was supposed to be released in 2012, and they got delayed by a number of years for refactor after refactor. And you had earlier said the hope this is like what maybe the original vision of Magento 2 should have been. What are you hoping that developers now take this, and what would you like back from developers when they're using the framework? Mark Shust (20:20.035) I would just like feedback, know, let me know what works. And it's funny you speak to Magento 2 being delayed year over year. I remember that it was like five or six years. It was a long time that kept getting delayed and pushed back. And what's crazy is I've had this kind of vision for how I wanted the framework to be, how I wanted it to work, full PHP, PHP versus in support with everything. No XML, simplified, you know, all these loud error messages by design tell you not only what's wrong, but how to fix it and why it's wrong, which is lines really well. Also, my teaching experience and just a very pragmatic language as well that just doesn't do a whole lot of magic behind the scenes. It's explicit and you know exactly what's going on. It's opinionated, but it also opens up, you know, avenues that you can customize however you want. And yeah, when I was building this whole thing, I actually built the whole thing with AI and cloud code. And I built the whole thing myself. I got the whole core architecture done in about 30 days. And it's, it completely changed workflow. So you go from what years with a team of I don't know how many they had on it, 10, 20 developers trying to get Magento 2 working. And I am a solo developer who's built something that's, I believe is more powerful, is simpler to learn and easier to modify. And I did it in the core of it in 30 days. Now it definitely isn't mature or experienced or anything. So it needs a lot of rigorous testing and feedback to make it even better. I've done my initial review of it and I think it's good enough to ship a 0.1 version. But I would love feedback from all users and let me know how I can make it even better. Brent Peterson (22:28.696) You know, I think one of the things that Adobe is trying to do is they are trying to make services out of Magento, out of Adobe Commerce. Do you feel as though this could serve a purpose in the bigger picture of what Magento is, Adobe Commerce is into the future as they seemingly start splitting up the parts of Magento into smaller pieces until they can, I don't know what their plan is, they have a new commerce. engine behind the scenes but do you feel as though this can really lend credence to what they're doing? Mark Shust (23:06.899) I don't know, they're taking every, they're kind of going the opposite route. So they're taking their framework and they're closed sourcing it, which I am not a fan of. I've been very vocal about that. I don't think your code should be in the hands of a SaaS, something like Shopify. When you lease out your IP to another company or you're giving so much responsibility to a third party that you have no control over, they can flip a switch and change things overnight and. you know, not give you heads up or they could be in your store or, it just, you have no flexibility when you're inside a SaaS. And I don't see why any of that would apply just to Shopify. It applies to Adobe commerce, SaaS, right? 100%. And, I'm, I've always been a huge proponent of open source. Marco's a hundred percent open source on an MIT platform. You can host it anywhere you want. download it for free and write from packages. You don't need a weird composer authentication process or anything like that, which makes it easier to contribute, easier to use. So I'm just, I'm all in on open source and I think that's the way to go. And I don't understand how you can innovate even in today's day with AI and lacking the ability to build stuff yourself and deferring that to a third party, it's not appealing me. Brent Peterson (24:38.03) I know from an app standpoint, you can build it on any framework you want for Shopify, let's say. This would be a great solution for even a Shopify developer to have to build, if they have to build an app to connect to Shopify, they could be using the Margo framework for this. Mark Shust (24:57.249) I absolutely you can. So Marco is just PHP behind the scenes and it's not it's not a very siloed in framework that you have to use it for commerce or it's only good for this thing. You can use it for anything. It's you can build what what you would normally build in a Laravel app. You could build in Marco. Same thing with Magento. Same thing with Drupal. You can even to tell you the power is it's again it's just a framework. So if you did really want to build a third party commerce system like Shopify, you can actually do it on top of the Marco framework. So you can build a multi-tenant platform on the Marco framework, get all of the plug and play advantages of modules that even, know, none of this is built out, but I can envision someone building a whole module marketplace ecosystem with the O. web GUI, kind of like what we had a long time in the past, where you can just click modules and install them and get everything to work. Of course, you could do it with vetted modules and ones that have been approved by you or your team, but you can build something like that on top of this framework. There's no reason not to, and you can host it yourself. have kind of an anti-SaaS view on using it, but for builders, you might want to build something like that for your own clients. that might be a request that you come in with a service job that you actually have to something like that. Brent Peterson (26:28.802) Yeah, and I think even the way Adobe is heading, they want you to use their app builder at Marketplace to be attaching to their different platforms that they're launching, including Adobe Commerce as a service. Adobe Commerce Cloud as a service, I think what they're calling it. You mentioned performance. Hufa came out a couple of years ago, five years ago now, geez, or more than that. that really increased the performance of Magento and then PHP has only helped in that. Where do you see Magento 2 going and the longevity of it and tell us, just give us your thoughts on Magento 2. Mark Shust (27:09.915) I, I've seen adoption for new projects drop, myself. there are tens of thousands of Magento stores out there that are heavily active and maintained. And I think they will continue to be on Magento for the foreseeable future. once you build a level of code and architecture that, that your whole business relies on, it tends to stop making sense for you to migrate to another platform. It's just. existential risk, even in the age of AI and mass advancement. You know, I think it does make sense to stay on those platforms. But I definitely seen new adoption drop. seems like most new e-commerce is going to Shopify. Unfortunately, that's just what's happening. It makes business sense, makes financial sense because time to market and costs is, you know, more crucial than anything. So, yeah, I see I've seen the market drop, unfortunately. I, again, my whole, it's crazy I'm saying this, because my whole full-time job is M Academy, right? It's making Magento content and courses. But yeah, I think Adobe can put forth the effort into actually innovating on the framework. But unfortunately, we haven't seen any of the innovation that I can even build myself in 30 days, right? with my own framework. there's, you know, when you have thousands of developers available, I don't know what the excuse winds up being. And I know it's very critical and probably hard for some Magento developers to hear, but that's exactly what's going on. So it's kind of part of the reason I was also building Marco too. It's like, I want to innovate. I want change. I want modern, modern updates. I want to be able to, you know, have the alum build out my code, a lot of it, and build it fast. I want full control over everything. And Adobe seems to be doing the exact opposite with everything with the Adobe Commerce software as a service. They're closing everything. We haven't seen much innovation in the ecosystem. So if we saw some new modules and new functionality and changes to the core or maybe a new admin interface that provides real value that's faster or a different way of doing things, right? Mark Shust (29:39.555) Can we have an alternative to XML? Like the big things, you can't be afraid of breaking stuff. So I think that's the type of changes we need to see in the platform and the ecosystem. And I just haven't seen it. So unfortunately, it's a little grim of the future, but hopefully, MageOS and HUVA and projects like that really definitely bring fresh blood into the ecosystem and drastically help out the whole community. I think projects like that are fantastic and we need more of them. Brent Peterson (30:13.462) the kind of the messaging I hear from the European crowd is that their adoption is growing and maybe part of it is that people are moving to the Hufa theme and it may not be net new build and I don't have data on net new build but I know that the market is very, the Magento market is still popular at least in the Netherlands and probably Germany. Is there some kind of messaging we can give to the Europeans that put a little bit more attention on the US market. I just feel like, I feel like that the Europeans who are now driving Magento are forgetting about the US market and how big it is. Mark Shust (30:55.395) I can't really speak on markets too much. I'm not too intelligent on them, be honest, the geographic locations. Whatever they're doing, I would love more of that in the States, of course. But any kind of innovation that they can supply to us or make widely known, that would be great. Brent Peterson (31:19.178) So your students, do you still have a lot of students coming out of India? And India is still just a huge development center for all things e-commerce. Mark Shust (31:27.851) it's always been. Yeah. It's pretty split. So, McAddy is very international. We have students registered in over 85 countries, over 15,000. and growing. Yeah. So, it is all over the world. It's so, but as far as percentages and where they're coming from, definitely a lot of activity in Europe, in the States, in India. So I do see it widespread, but I have seen adoption tick down myself. So that's just for my vantage point. Now hopefully Europe is different. Brent Peterson (32:10.574) Anything that you want people to know about Marco, any way that they can get it. Tell us the best way for them to download it, to try it, to play with it, to build resources, to do some community around it. Mark Shust (32:26.627) Sure, check out, I have tons of posts about it. Just check out Marco framework. You can go to marco.build. That's a publicly facing site, which will have links where you can use Marco talk. You can connect with other developers that are using it. You can check out the GitHub from there and get a rundown of kind of what makes it tick and why it's important in today's. Tay's world. So that will give you a lot of info on it. And of course, the readme and the GitHub, it has extensive documentation. I was very adamant about what does this do? What does that do? How do I build something? You know, what do I have to do to implement a specific feature? It's all fully documented. Right now it's over, I think, 17,000 lines of and it's extensively documented. yeah, I'd love any feedback you have. It's going to be very open framework. So be open to taking in PRs and really innovating and growing with the larger PHP ecosystem. Brent Peterson (33:40.182) Any challenges you want to issue to some developers out there and what they what you'd like to see done on the framework? Mark Shust (33:46.431) Of course, the commerce, you know, that's number one. I want to see a commerce package. I will be building one. I'm sure myself, but I like you to push it. All right. Push the limits. The server side event module is a great idea and it's you can definitely install the Marco framework locally. I will have development workflows in place. They're very advanced where Brent Peterson (33:59.567) you Mark Shust (34:13.239) My typical development now is I talk with the AI for about 10 or 15 minutes and flush out a full plan. And then I have it completely carry out the plan autonomously. It has a plan writer, a devil's advocate reviewer. Once you approve the plan, it kicks off autonomously on a loop with test-driven development. So everything is tests. Everything is extensively documented. It follows code patterns and standards. You'll probably still have issues, of course, when you're when you're it's done building, but it they should be few and far between. I've had very good experience like building the server side events modules and Redis modules and things like that. So, yeah, it's anything you think would make a great module for Marco and that could be used across tens of thousands or millions of sites. Definitely try to get it in the framework so you can use it as that core piece and then build on top of it for your own custom functionality. Brent Peterson (35:17.934) I should ask, is it is Claude, Codex, Gemini, any of the builders out there that you'll be able to glom onto? Are you favoring one or the other? Mark Shust (35:28.643) Oh, I've been, I've been using cloud code since the day it came out and I have not used really anything since it's been worked. It's worked great from day one. I have a course on it. Of course, I have an Academy. If you want to check it out, I'll teach you all about it. But, um, I am a huge cloud code fan. Uh, I like, it has a lot of very specific ways to do things, uh, which I like how it's opinionated like that. Uh, and it, uh, It just allows me to do everything that I need to do. I haven't had any reason to check out another platform. It's just, the models are so good and they're innovating and they're advancing and making it better all the time. So yeah, that's what I've, but that said, of course I'd love it to work with every other, know, LLM there is that's out there and I see no reason why it can't do that. Brent Peterson (36:21.186) That's awesome. Shust is m.academy and now marco.build. What a great conversation. Thank you. Mark Shust (36:30.308) Thanks a bunch. Thanks for having me, Brent. Brent Peterson (36:34.937) Okay.