I use haproxy just for simplicity, with the https fronted in tcp mode with a tcp request inspect delay and acls using req_ssl_sni allows switching between multiple backend based on hostname, though if you only had 1 backend server then it could just be a raw tcp proxy. I have my stuff split between 2 local boxes that are both connected to the reverse proxy using wg so this works well for me.

I use a wireguard tunnel as a reverse proxy and it does ssl passthrough with sni. Termination happens on my local box and the reverse proxy host doesn’t do any certs.

I went with rsync.net, grabbed a promo at some point and have been getting 1.6TB for $10/mo. They support a wide range of protocols (sshfs, sftp, rsync, borg, synology, guides for freenas and truenas), snapshots, idle alerts, geo redundancy for a fee. Been very happy with it so far. Had to retrieve some files once and had no issue.

I couldn’t find a common compatible format for me and 15 friends :( I also got tired of getting 20 requests to download more movies and episodes every week.

Looks like it’s a feature of the router they provide you if you don’t bring your own.

Removed by mod

I do Bandcamp where I can but end up just yt-dlping YouTube and SoundCloud where necessary, and then I try to ask those artists to put their music in bandcamp or somewhere I can buy a download.

Doesn’t this need an AI disclosure in the title and post per the rules of this community?

Personally I would love a browser extension that detects these and adds a badge to AI content.

I emailed them at support@greenfi.com with basically the message “I can’t login I get this message, what can I do to resolve this”, went back and forth a bunch, did all the annoying troubleshooting reinstalling the app, clearing app data, etc, sent screenshots of the message, assured I wasn’t using a VPN, sent all the details of my phone and rom version, had to poke them when they didn’t respond, got a random message of them saying “we see you were able to gain access” and I had to be like wtf no I didn’t?, told them it worked if I temporarily enabled play protect and said I would switch banks if I couldnt access check deposit which is mobile only, they replied with some justification why play protect is tripping it up, so I told them that I had no issues having play protect off with other financial apps like fidelity and chase bank, and finally got a response from a “supervisor” saying “i refreshed your settings” settings" and it started working.

Not sure if they just whitelisted my account or what, but maybe if enough people complain they will get tired of whitelisting everyone and change course.

I switched away from Ally bank for similar reasons, on my lineage’d pixel 5 with gapps, their app would always fail the login the first time then succeed the second time for no reason.

Then the bank I switched to, greenfi (coastal community bank) completely blocked login unless I had play protect (in Google play) enabled, which allows Google to uninstall any app from your phone without asking if they decide it’s a “security risk” or whatever - I can totally imaging them deciding that ReVanced / morphie / patched apps that block their ads to be a “security issue”. Luckily greenfi was more understanding and they did something that allowed me to log in again.

I completed a 300 mile move yesterday, and my homelab was one of the least stressful parts thanks to my pre-emptive downsizing from a half rack to just a Synology + Intel nuc. I also pre-emptively switched from forwarded ports to a reverse proxy because I didn’t want to be thinking about what isp I would end up with. Even further, I linked the nas and nuc directly with Ethernet so that I wouldn’t need to rely on static IPs or router-assisted local hostnames, mdns, avahi, whatever because I wanted to be able to transplant my lab onto any network just in case of emergency - thankfully I haven’t had to test that out but it’s nice knowing my lab should work on any network where both boxes can reach the internet / my reverse proxy. I was thinking worst case if we ended up in some apartment complex with some crappy provided setup that had broken local discovery then it would be 1 less thing to worry about.

Fyi the filters are instance based, so the person you replied to and probably some of the people who downvoted you have no idea what you’re talking about. Only we can see it because we’re on .ml

This is not for the case of using text, it’s for protecting text in images. The problem with adding invisible text is that LLMs cam defeat it by screenshotting and using OCR. This tries to plug that loophole.

Invisible text doesn’t work for images though, which is what this is specific for. Invisible text is also defeated by taking a screenshot, so if you use this too then you have a chance of defeating both LLMs that read the text as well as those that screenshot and OCR.

For some reason the battery isn’t mentioned in the parts list but it’s shown in the video and mentioned in the instructions. Would be pretty easy to power the esp feather with a dcdc converter off of a USB port or something assuming it doesn’t take 5v directly

To re-state, it’s not a problem of understanding or the language, it’s a problem with overused, bland, and repetitive language and phrasing. It’s not fun to read even if you can understand it perfectly.

Just because 200 people understood your post doesn’t mean it isn’t annoying to read. And it’s a super simple thing to fix - don’t use AI to rewrite your post and just use a normal translator instead, which won’t inject bland and repetitive phrases and structures.

Yeah Google translate preserves a lot more of your communication style. AI overuses so many phrases it can get annoying reading something that was processed through it.

It feels like there’s been an increasing flood of AI slop projects, with varying degrees of monetization / donations. I think it’ll become a huge problem if we don’t have at a minimum very strict rules around AI generated slop projects. I think a mandatory tag with penalty of removal is a good bottom floor, in addition to the recent community participation activity % requirements for promoting monetized projects, which covers a good chunk of AI projects.

Then hopefully soon we can figure out as a community what to do to control the remaining volume of non-strongly-monetized AI slop projects if those are still too widespread, but having it labeled is absolutely needed transparency, and that’ll still be difficult because lots of people seem to lie about not using AI.