Show HN: Engauge – Self-hosted, Single-binary Analytics | Hacker News Reader

Show HN: Engauge – Self-hosted, Single-binary Analytics

github.com

38 pointsengaugeai27 comments

> We apologize for having to do this, but Engauge is closed to code contributions (at this time). This keeps the code base free of proprietary or licensed code.

I also see project license is called PolyForm which I’ve never heard before. Is this an open source license? If not, you may want to change open source section title to avoid confusion. Also, could you explain what is the pros and cons of this license briefly?

The Competition and Noncompete sections of the license aren’t inline with open source licenses. It’s more of a SSPL style license.

The license as a whole feels a bit like a random hack job.

This project should call itself source available.

The non-compete specifically is dangerously under-specified. This is close to crayon license issues https://opensource.stackexchange.com/questions/1445/how-can-...

Open source != free for all

I think it's good that people use alternative licenses. It's about time the situation, where there is a nice project conceived in a garage and developed into maturity with the hands of volunteers and then appropriated by a big corporation that then make billions out of it without sharing any profits, ends. Open source was never meant to be exploited this way, but big corporations started to see it as a big R&D money saver and started promoting it. We should see through that. Developers deserve to be paid and shouldn't be exploited by big corporations who wipe their conscience with Open Source PR.

Yes, great questions!

Polyform is an awesome set of open-source licenses: https://polyformproject.org/licenses/

The choice of license is perhaps ambitious / premature. It is chosen so that I can add whatever features to the open-source version being confident that some startup (or possible even larger company) could not attempt to productize it themselves.

> PolyForm is not…

> Open source or free software. There are plenty of existing open source licenses. PolyForm is not a substitute for them, but an alternative for those who want to license source code under limited rights.

https://polyformproject.org/what-is-polyform/

Ah yes, I stand corrected. It is difficult to discuss these things sometimes as everyone has a slightly modified definition of "open-source", etc.

I guess I did not see a valid reason why this project would be used as a dependency in another software project/product (open-source or otherwise) so Polyform did not seem limiting.

I understand your concern. Open source licenses are the ones that Open Source Initiative approved and just a note, you can’t have an open source license and prevent someone else to create a product from your work. Anyway, hope you find a license that fits your requirements. Good luck with your project.

I'm not an open-source purist myself, and Polyform seems like it might be very useful for many kinds of projects.

Just be careful because some people get very salty about things claiming to be "open" when what they mean is "source-available".

I added your solution to my list of awesome web/app analytics solutions:

https://github.com/underlines/awesome-marketing-datascience/...

Thanks! The list looks great! :)

I wish people would realize using pointers in go doesn't automatically mean better performance. You should really stick to using values and only use pointers if it's necessary/makes for better code/increases performance. Using pointers makes escape analysis harder and results in more pressure on the gc which can effects performance more than usage of pointers saves.

What database does it use? How well does it handle high traffic and filtering/aggregation of data?

To avoid dependencies it utilizes the filesystem for now storing raw data in CSV files but storing analytics (the information in the dashboard) as gob files.

This is part of why it is simpler to deploy. Just build and run without setting up a db.

> How well does it handle high traffic

It seems to handle things quite well. I couldn't give you exact figures, but I have done local low 4-figure rps testing and it didn't sweat all that much.

A click on the "db" folder will answer the question. It's storing the data in local csv files. It's neither crash proof (can corrupt files) nor usable for high traffic or bigger amounts of data. They hint at an "enterprise" version using a proper database...

What differs your product from products like Matomo (besides being a single-binary)?

Matomo is a pretty high standard of comparison, haha. So, thanks! Long term plans is to turn this into a self-hosted, single-binary Firebase alternative of some sorts.

For example, current features in testing are things like funnels, AB/N-Tests, etc. Then extend with any features that allow direct engagement actions for active users.

EDIT: another benefit might be performance of Go over PHP. But that may only show if you plan on hammering the service.

Don‘t get me wrong, I don‘t want to criticize you. I love a good piece of open-source software and I absolutley love doing something for fun just because I can (and not because there‘s a need/market for it). One can learn something at all times.

However, Matomo has quite „ok“ server requirements when tracking ~100 million page views per month [1] - and it’s in PHP... Go will most definetly still perform better. And maybe it just felt like it, but I think there‘s just the other week a bunch of tools doing web analytics shown here. So my first thought was „and another?“.

So I hope you‘ll get something of it and you gain some momentun, competition is always good!

[1] https://matomo.org/docs/requirements/#tracking-more-than-100...

Matomo will probably run circles around this tool because it actually stores data in a structured format that makes is optimized. This software might grind to a halt pretty quickly. Being written in Go is all fine and will help with handling the http requests but the data handling seems inefficient here. It stores data in local csv files that it reads and decodes in whole. There is a lot more work to be done before this can be considered to be used in production imho.

It is true that the raw data is stored in append-only CSV files. But the dashboard data / analytics data is stored in gob binary files. Very small and efficient.

The architecture is stream-based analytics instead of query-based.

why not just use google analytics and be done with it?

Great question! This allows you to control your own data for later custom analysis if you would like. It also allows you to track custom metrics / custom data from the backend so that it cannot be spammed.

If I’d gone to the homepage rather than the repo I’d have answered my own question. Cool stuff, best of luck!

Maybe next time spend a second to think about why anybody would like to do that — could be that you find some very valid and good reasons you never thought about. No need to agree to all of it, but the ability to put yourself into other perspectives is a much needed resource these days.

Google e.g. is a US company. US companies have to comply with the patriot act, meaning they have to help US secret services to access data Google collected. By doing so they would however be in direct violation of GDPR. This means from within the EU technically you should not use Google Analytics. (This has afaik to do with the ruling of the European Court of Justice, where a law which declared US privacy standards to be in alignment with EU privacy law was struck down).

And neither do I want to use Google Analytics. On my websites my visitors trust me with their data. This means it is my duty to understand what happens to their data on my website: what is stored, what is forgotten, what is sent of to which third party and for what reason. Google Analytics makes it kinda hard to understand that, so I'd rather use something that is simpler and stays on my servers and within the legal framework of my country.

> On my websites my visitors trust me with their data.

Really? most customers don't really care.

I highly tune my Google Analytics to be as fast as possible and we get great insights from it, works great for me and also GDPR compliant to as I ask for the users permission.

Why not host yourself if you can? Why feed Google?

Don't have the time to host myself, much rather use something already built and battle tested.

Reply on news.ycombinator.com