Luke Boswell (Jul 28 2026 at 05:38): Luke Boswell (Jul 28 2026 at 05:38):Just wanted to share that I've updated my htmx-playground example to help me test the latest release of basic-webserver
https://github.com/lukewilliamboswell/roc-htmx-playground/
It's a much more realistic web app and I've spent some time trying to explore different features like static dispatch and how that might look for this use case.
I've used a lot of AI to generate things, but my goal here is to factor and structure everything for maximum compile-time type safety and testability as things grow.
There are definitely rough edges to polish in Roc and bugs I found along the way. I've been able to work around those ok.
If you check it out, I'm keen to know what you think.
bitbegin (Jul 28 2026 at 15:56):Can it be packaged into an exe?
Anton (Jul 28 2026 at 17:39):Yes, if you execute roc build main.roc on windows it will produce an exe.
The exe does not contain everything though, it expects an assets folder as the app is currently written.
Anton (Jul 28 2026 at 17:40):Yes, if you execute
roc build main.rocon windows it will produce an exe.
Perhaps not right now, because I noticed a specialization issue (on macOS) that is eating all the ram after which the process is killed.
Clarification: this specialization issue does not affect all programs, I've only seen it with roc-htmx-playground so far.
Luke Boswell (Jul 28 2026 at 21:25):Curious.. Ive been running it on Macos without any issues, thats my dev machine. I might be on an older commit, I keep switching around while Im working on roc bugs concurrently
Luke Boswell (Jul 30 2026 at 23:31):I hope folks here dont mind... but I've decided to pivot direction with this. I'm not sure how many people were looking at the playground anyway.
It started out as an experiment, and I've been on a roll lately cleaning up older repos to migrate to the new compiler or retire if I'm not wanting to maintain it.
I think the general roc+htmx+webserver example has genuine value to people, and I don't want to take it away or anything, but in it current (well previous now) form it wasn't something I really needed or would use and improve.
So, I've decided to convert the playground into a real app, specifically a simple CRM. I've been doing a lot of business development lately and needed a real CRM anyway.
I know the general notes about not rolling your own, but I've got many reasons why I think this is the right approach for mu business. Most significantly it's a useful tool for me to talk about specific engineering and business trade-offs -- without sharing IP from my clients' projects.
Folks are welcome to use it, share ideas or provide any feedback as always. I've got a particular scope and priorities in mind for the roadmap so I can showcase certain things.
Bryce Miller (Jul 31 2026 at 00:24):As someone who customizes a commercial ERP product for a living, I think people generally underestimate the costs of buying + customizing a solution.
Luke Boswell (Jul 31 2026 at 00:26):I looked at the various things around, and I thought if I'm going to spend significant time working on something I'd really it to be written in Roc.
Bryce Miller (Jul 31 2026 at 00:29):Yeah, imo a canned solution might take you from 0 to 100 faster than something custom, but then going from 100 to 500 will be substantially more costly.
Bryce Miller (Jul 31 2026 at 00:31):Plus, if you build it youself you have full freedom to fix any problems you create.
Bryce Miller (Jul 31 2026 at 00:31):So all that to say, this doesn’t actually sound crazy to me
Aurélien Geron (Jul 31 2026 at 00:47):Yeah, I'm also looking into CRMs these days (actually even full blown ERPs) and I haven't found anything I love. Frappe is the least bad so far. Given the capabilities of coding agents, it might be better to roll out my own too.
Luke Boswell (Jul 31 2026 at 00:49):You're welcome to experiment with this one if you like... it's quite primitive right now but give me a few days and I'm sure it will be much nicer to use
Luke Boswell (Jul 31 2026 at 00:51):I got bored typing in people details so the first feature I'm attempting is automating the business card to new person details workflow. I want to take a photo using my phone and have a cheap agent extract the key details and prefill the form for me.
Luke Boswell (Jul 31 2026 at 00:52):I guess I'm lazy like that... :smile:
Luke Boswell (Jul 31 2026 at 00:54):Frappe https://github.com/frappe/crm looks nice
Luke Boswell (Jul 31 2026 at 01:33):I haven't built something like this before... but damn I am impressed with how cheap and accurate these things are! This is one image.
Aurélien Geron (Jul 31 2026 at 01:36):Be careful as you scale: once you have 1,000 new customers per month it will cost you 64 cents/month. :sweat_smile:
Luke Boswell (Jul 31 2026 at 01:37):Wow ... I'd love to have that problem!
Last updated: Aug 12 2026 at 12:35 UTC