The most bleeding edge stack of all time
August 3, 2024 8:37 AM Subscribe
How to save $13.27 on your sAAs bill. Hilarious. Sometimes, the boundary between the real and the parody is just a few lines of code.
"$14 per 100k events", SAAS is such an incredible racket.
posted by mhoye at 9:57 AM on August 3 [1 favorite]
posted by mhoye at 9:57 AM on August 3 [1 favorite]
SaaS : hidden operational sunk cost fallacies :: Clicker Games : dopamine
posted by Lenie Clarke at 11:51 AM on August 3
posted by Lenie Clarke at 11:51 AM on August 3
I got to "try and save a few $ on my Vercy bill by building an analytics api from scratch using a new stack" and it reminded me of having fun with all this internet stuff.
Fun. Sqlite amusements. The api got chosen because one "has a delicious looking mascot" and the other because it has a funny name. No daily backups, snapshots, point in time recovery. Love it. Delicious.
But then. "rookie mistake of having skill issues with docker".... and god. Good old just stick the thing in a box and then all your problems are contained. wooo. And, if you want, just for giggles - you can make copies and copies. and copies. Because he problems I am solving are totally stuff that might need to be scaled all the way to INFINITE BOXES. And I have to say I was particularly entertainment by the RSC shenanigans. It truly is it the world we live in.
The "we" I'm talking about is modern web developers. Web analytics is not something that anyone really thinks about anymore - it's a 'solved' problem. You can just pop in a little chunk of google code, rely on whatever free junk bin parts your web host provides or pay for levels of 'pro' data.
Sqlite is a whole mini database. It's not total overkill. Learning two whole new javascript API things? Chasing the new hotness, just on a whim? All that effort just to use them on one project? What isn't stated is that all this effort will just be abandoned for the next project. And then a testing regime that is disconnected from reality? And that's all before the React Server Component circus, which doesn't even get named beyond its acronym and upon which several different chart libraries are smashed.
posted by zenon at 1:00 PM on August 3 [4 favorites]
Fun. Sqlite amusements. The api got chosen because one "has a delicious looking mascot" and the other because it has a funny name. No daily backups, snapshots, point in time recovery. Love it. Delicious.
But then. "rookie mistake of having skill issues with docker".... and god. Good old just stick the thing in a box and then all your problems are contained. wooo. And, if you want, just for giggles - you can make copies and copies. and copies. Because he problems I am solving are totally stuff that might need to be scaled all the way to INFINITE BOXES. And I have to say I was particularly entertainment by the RSC shenanigans. It truly is it the world we live in.
The "we" I'm talking about is modern web developers. Web analytics is not something that anyone really thinks about anymore - it's a 'solved' problem. You can just pop in a little chunk of google code, rely on whatever free junk bin parts your web host provides or pay for levels of 'pro' data.
Sqlite is a whole mini database. It's not total overkill. Learning two whole new javascript API things? Chasing the new hotness, just on a whim? All that effort just to use them on one project? What isn't stated is that all this effort will just be abandoned for the next project. And then a testing regime that is disconnected from reality? And that's all before the React Server Component circus, which doesn't even get named beyond its acronym and upon which several different chart libraries are smashed.
posted by zenon at 1:00 PM on August 3 [4 favorites]
> You can just pop in a little chunk of google code, rely on whatever free junk bin parts your web host provides or pay for levels of 'pro' data.
i speak as someone with little familiarity with the field in the modern sense: doesn’t this result in google harvesting very valuable data / having an even more disturbingly panopticonic view of the entire web than they otherwise would have?
like i may be wrong on how things work or else people may be entirely blasé about the existence/extension of the google panopticon, but nevertheless a real bad thing that everyone’s become accustomed to is nevertheless a real bad thing.
posted by bombastic lowercase pronouncements at 1:14 PM on August 3 [5 favorites]
i speak as someone with little familiarity with the field in the modern sense: doesn’t this result in google harvesting very valuable data / having an even more disturbingly panopticonic view of the entire web than they otherwise would have?
like i may be wrong on how things work or else people may be entirely blasé about the existence/extension of the google panopticon, but nevertheless a real bad thing that everyone’s become accustomed to is nevertheless a real bad thing.
posted by bombastic lowercase pronouncements at 1:14 PM on August 3 [5 favorites]
I think we lost something from the dialup days, of websites optimized to load quickly over 14.4kbps connections. Everything is so bloated these days, with hundreds or thousands of cross-site scripts and trackers. I hate it here. :( Don't get me wrong, there's so much more to do online these days, but gods above, I'm sick of the bloat.
posted by xedrik at 1:14 PM on August 3 [8 favorites]
posted by xedrik at 1:14 PM on August 3 [8 favorites]
Almost 80% of websites embrace the google panopticonic with google analytics.
posted by zenon at 2:22 PM on August 3 [3 favorites]
posted by zenon at 2:22 PM on August 3 [3 favorites]
I've been enjoying using Goatcounter which is a single Go binary and a sqlite database, capturing useful data but not too much or too intrusively. Not as fun as rolling your own this way but built in the same spirit.
posted by i_am_joe's_spleen at 2:25 PM on August 3 [3 favorites]
posted by i_am_joe's_spleen at 2:25 PM on August 3 [3 favorites]
> Almost 80% of websites embrace the google panopticonic with google analytics.
precisely.
posted by bombastic lowercase pronouncements at 3:42 PM on August 3
precisely.
posted by bombastic lowercase pronouncements at 3:42 PM on August 3
doesn’t this result in google harvesting very valuable data / having an even more disturbingly panopticonic view of the entire web than they otherwise would have?
a real bad thing that everyone’s become accustomed to is nevertheless a real bad thing.
"sure but we already buy adsense ads so what's a little more data, and then it's all integrated for the quarterly numbers?"
i... wish i were kidding. this has been your depressingly-accurate lowercase reality check for the day
posted by 1xdevnet at 3:43 PM on August 3 [5 favorites]
a real bad thing that everyone’s become accustomed to is nevertheless a real bad thing.
"sure but we already buy adsense ads so what's a little more data, and then it's all integrated for the quarterly numbers?"
i... wish i were kidding. this has been your depressingly-accurate lowercase reality check for the day
posted by 1xdevnet at 3:43 PM on August 3 [5 favorites]
I really do wish I were kidding. I've tried pushing self-hosted analytics (as well as "you don't really need analytics for this") and either got a variation of that, or "but who will maintain it?" Goatcounter seems nice, though.
posted by 1xdevnet at 3:47 PM on August 3
posted by 1xdevnet at 3:47 PM on August 3
As someone who has implemented something similar for my very low traffic blog, I can appreciate the work that went into this project and the desire to save some money. I appreciate the comments about not giving all that data to google, but really websites should be going a step further and not collecting data such as ip addresses at all.
The GDPR is very specific about what you are allow to keep, and logging peoples IP addresses forever for no good reason is definitely out. That is why my solution just maintains a simple hit count.
That is the end of my rant on the subject.
posted by AndrewStephens at 4:42 PM on August 3 [3 favorites]
The GDPR is very specific about what you are allow to keep, and logging peoples IP addresses forever for no good reason is definitely out. That is why my solution just maintains a simple hit count.
That is the end of my rant on the subject.
posted by AndrewStephens at 4:42 PM on August 3 [3 favorites]
Reminds me to go clear logs… how much effort is requested to modify existing Apache to NOT log IP addresses? And almost all IP’s are pathetic attacks on windows software on a Unix vm.
posted by sammyo at 4:53 PM on August 3 [1 favorite]
posted by sammyo at 4:53 PM on August 3 [1 favorite]
Hono because I saw this video where a guy said Hono and it made me laugh
Hello, me.
posted by Literaryhero at 6:01 PM on August 3 [1 favorite]
Hello, me.
posted by Literaryhero at 6:01 PM on August 3 [1 favorite]
Does make me wonder how much $$$ I've saved by just using analog on apache.log for the past 30 years.
posted by joeyh at 6:20 PM on August 3 [2 favorites]
posted by joeyh at 6:20 PM on August 3 [2 favorites]
TIL "Gypity" = ChatGPT
posted by thecjm at 8:07 PM on August 3 [4 favorites]
posted by thecjm at 8:07 PM on August 3 [4 favorites]
how much effort is requested to modify existing Apache to NOT log IP addresses?
Very little, just read the docs for the LogFormat directive and you can define whatever you want to be logged or not logged.
Does make me wonder how much $$$ I've saved by just using analog on apache.log for the past 30 years.
Heh. I am just surprised there's still an Ubuntu and a Debian package for it - I don't think the author has updated it in many years. I used to love analog (and the original author's insistence that web analytics are bullshit and he threw analog together to placate his employers). Maybe I should have another blat with it.
posted by i_am_joe's_spleen at 1:12 AM on August 4 [2 favorites]
Very little, just read the docs for the LogFormat directive and you can define whatever you want to be logged or not logged.
Does make me wonder how much $$$ I've saved by just using analog on apache.log for the past 30 years.
Heh. I am just surprised there's still an Ubuntu and a Debian package for it - I don't think the author has updated it in many years. I used to love analog (and the original author's insistence that web analytics are bullshit and he threw analog together to placate his employers). Maybe I should have another blat with it.
posted by i_am_joe's_spleen at 1:12 AM on August 4 [2 favorites]
For Debian package maintainers, lack of updates just makes packaging easier. Ideally the package has no users either. Upstream maintainers and downstream users just generate bugs and bug reports. In their absence, you are freed to concentrate on the package itself as a pure, crystalline object. (Tongue in cheek; I am a former Debian developer.)
posted by mbrubeck at 8:56 AM on August 4 [4 favorites]
posted by mbrubeck at 8:56 AM on August 4 [4 favorites]
« Older The very serious function of racism | Jax wrote a song for her dad Newer »
posted by mbrubeck at 9:12 AM on August 3 [4 favorites]