First off; I should note that this is 100% shameless self-promotion. Its also an excuse to finally stop lurking here like I have for the past 3 years.
I just started working for a brand-new company called Earthly. We make (in my very biased opinion) a pretty slick build tool that makes builds/tests/whatever run the same locally and on your CI machine, among other things. A one sentence pitch would be: “It’s like a Dockerfile and a Makefile had a baby!”
We’ve been trying to do some outreach to get people using our new tool. Its free, and source available (BUSL 1.1). Reaching out to open-source developers directly isn’t quite bearing a ton of fruit; turns out cold-pitches aren’t super effective.
So here’s a deal I am willing to offer: I will improve/create your open-source CI using Earthly, if you want. Or, if you know of any open-source projects that who might be interested, would love some pointers. Heck, if you have a closed-source system I would be willing to do it too, since I just want people to try this thing we’ve made.
If you don’t have a project and try it out or have feedback, that is always welcome too. The continued existence of my job depends on this thing improving and catching on
I have several (most?) of my projects using GitLab CI and I love it. On most projects it handles testing and on my professional projects it handles deployment as well.
Say I were make the switch, what benefits would I get from using Earthly rather than something like Travis or something else?
Last, what is your incentive to spread the good word? Nothing wrong with being paid to make a sales pitch, but I want to make sure we know where we stand.
I have several (most?) of my projects using GitLab CI and I love it.
Keep using GitLab CI. We have one user who we actually helped get up in GitLab just recently! Think of us more like a layer of indirection between you and your CI.
The guy using Earthly + GitLab Runners just has GitLab configure some docker-in-docker stuff, and also some general prep work. Then it invokes our tool. What was cool here is that when he had questions about why something was broken, I was literally able to clone his repository and run his build too to check what was up; got the exact same results as his CI (which I have no clue how it is configured… also don’t really have access to it).
what benefits would I get from using Earthly rather than something like Travis or something else?
Reproducibility. You can get the exact same thing to happen in GitLab and on your machine. We can also pop a debugger to let you inspect state. This also eliminates local dev machine setup/tooling (see asdf, nvm, etc)
Caching. We do some cool caching stuff that can let you share the cache of your CI to speed up local builds. This also helps in cases like GitHub actions where cache across runs isn’t really a thing.
You can share build parts easily without becoming a YAML developer - you can even use pieces of builds in different repos using a URL scheme.
I think our syntax is more readable and useable than the typical multi-stage Dockerfiles.
We handle multi-platfom builds very easily, using a single file rather than multiple Dockerfiles.
Last, what is your incentive to spread the good word?
I work for the company. CI and containers are super fun to me and I would like to keep doing it for as many years as I can, and that is 100% dependent on user acquisition leading up to more funding. Additionally, we are trying to push adoption so we can get real-world feedback. This will make it so we can build things that are legitimately useful; rather than guesses.
No pressure if you try it and hate it, I would just love to know why and if you had any suggestions.
Last edit, I promise: The CLI tool is currently free, and will continue to be free (BUSL 1.1, like what MariaDB is doing). Our plans are to monetize elsewhere.
Ok, one more: I am not a sales guy. I am an engineer. Am I doing this “developer evangelism” right?
I’m going to have a look at this when I’m finished with other responsibilities today and I’ll let you know what I think of it.
I admit, I’m a bit skeptical, because from what I can tell, it purports to be a magic solution to unifying all the CI configuration in a project? Seems too good to be true, tbh. but I’ll give it a fair shake a bit later.
Yeah me too. If it falls short, we will try to fix it.
purports to be a magic solution to unifying all the CI configuration in a project
This is the hope. I like to think of it as the “wire closet” for your CI/local scripts/integration testing stuff. We still have some ground to cover, but were trying! For instance, we don’t have a clean, documented solution for running in K8s yet (which is what I am slacking off from working on right now).
Seems too good to be true, tbh.
Fair, and probably a little true. But I hope we are closer to the claims rather than complete and total failure?
So I’ve been a bit more busy than I thought I’d be, however, I’ve had a chance to give it a once over.
I think it’s a really cool solution, and it’s definitely something I could see myself using in the future. Since I haven’t had a chance to really dig into it, I can’t yet say anything with regards to criticism. It seems like a solution that doesn’t necessarily cut down on the amount of configuration, but definitely puts it all in one place, which is nice.
I’m curious… how. From what I could tell, there’s no enterprise services solution offering on the site or any form of commercial add-ons or services. How does more users not giving you money lead to profit?
I ask this primarily because if you don’t have an existing funding sponsorship base OR some mechanism to fund yourself through services, in all likelihood the staying power just isn’t there.
While I’m intrigued by the product, this still looks like it has a decent amount of time investment required to implement and therefore is hard to consider if there’s a strong chance it won’t be around in 18 months.
I ask this primarily because if you don’t have an existing funding sponsorship base OR some mechanism to fund yourself through services, in all likelihood the staying power just isn’t there.
We’re seed round right now, and have some institutional backing. We do have plans to monetize… but as an open source style project, we need users before that plan becomes viable.
18 months
We have a little more runway than that, with optimistic visions for longer time horizons.
BTW; got any open source projects (or other projects) you would be willing to try this out? Any friends with projects?
