fmstrat

joined 2 years ago
MODERATOR OF
[–] fmstrat@lemmy.nowsci.com 1 points 2 weeks ago

Ahh, just realized the Mautrix Signal bridge I use no longer requires SignalD

[–] fmstrat@lemmy.nowsci.com 3 points 2 weeks ago

An unfathomably long list of projects and things to learn.

I like to pick up hobbies that require me to learn skills. Like sure, you could buy a 3D printer, orrrr you could build a Voron, learn FreeCAD, learn to filament tune, add a DIY air filter, add a DIY filament changer, and on and on.

[–] fmstrat@lemmy.nowsci.com 2 points 2 weeks ago

Wrong alert. It's multiple scans of an old photograph.

[–] fmstrat@lemmy.nowsci.com 14 points 2 weeks ago (5 children)

The author points out systemic issues with Flathub, why it needs to improve, and then, through some crazy rationale, still recommends it because:

Although Fedora Workstation works well for most users, and although quality and reliability has improved considerably over the past decade, it is still far too easy for inexperienced users to break the operating system

Without stating "how" it is too easy for this to occur. Flathub is not a "fix", it's a host of other problems, many of which are in the article itself.

[–] fmstrat@lemmy.nowsci.com 4 points 2 weeks ago

At room temperature methane is lighter than the atmosphere on Earth's surface, not heavier.

However, pressurized in your body it may he heavier. I see no studies measuring this.

[–] fmstrat@lemmy.nowsci.com 3 points 2 weeks ago* (last edited 2 weeks ago)

His wife told News 12 Long Island in a recorded interview that she was undergoing an MRI on her knee when she asked the technician to get her husband to help her get off the table. She said she called out to him.

Where does it say he ran in? I mean, what you say sounds right, but this doesn't read like "freaking out"

Edit: Sounds like she did not freak out, but called to him to help her stand up after it was complete (bad knee), but before he was authorized to enter. This seems more like an honest mistake and tragedy. https://www.cnn.com/2025/07/20/health/mri-machine-death-long-island

[–] fmstrat@lemmy.nowsci.com 1 points 2 weeks ago (2 children)

How is signald broken? I'm using it with Matrix now just fine.

[–] fmstrat@lemmy.nowsci.com 1 points 2 weeks ago

Is there a working Signal+Prosody setup?

[–] fmstrat@lemmy.nowsci.com 3 points 2 weeks ago

The problem is interpretation. You read it one way because of your life context and frame of mind. Many others will have a different interpretation.

[–] fmstrat@lemmy.nowsci.com 4 points 2 weeks ago (1 children)

Yes they are called "transports".

As for when I tried to change the URL, I was moving all services from publicly resolvable DNS to .lan, and Matrix is an issue because you can't change that after the fact, even if you haven't federated.

[–] fmstrat@lemmy.nowsci.com 2 points 2 weeks ago (1 children)

Doesn't matter bridge work by "adding a user" to rooms, vs just your user?

[–] fmstrat@lemmy.nowsci.com 2 points 2 weeks ago

Promising, would need to supplement GV and Signal, but this gets me halfway there, thanks.

 

Why YSK: Locking your credit with the main 3 places is not enough.

https://innovis.com/ have grown enough to require locking as providers are using them as a single check source.

There's also https://www.chexsystems.com/ which many banks use for opening checking accounts. They're unique because they handle accounts that don't show up in a credit report.

 

All the posts about Reddit blocking everyone except Google and Brave got me thinking: What if SearNGX was federated? I.E. when data is retrieved via a providers API, that data is then federated to all other instances.

It would spread the API load out amongst instances, removing the API bottlenecks that come from search providers.

It would allow for more anonymous search, since users could cycle between instances and get the same results.

Geographic bias would be a thing of the past.

Other than ActivityPub overhead and storage, which could be reduced by federating text-only content, I fail to see any downside.

Thoughts?

140
My favorite button (lemmy.nowsci.com)
submitted 1 year ago* (last edited 1 year ago) by fmstrat@lemmy.nowsci.com to c/foss@beehaw.org
 

It removes pay walls.

It rejects cookie requests.

It blocks ads.

It fixes the internet.

It is magnificent.

 

The Firefox for Android GitHub releases aren't being updated anymore, and is versions behind the Play Store now.

FDroid Fennec isn't fully the same app, and FDroid has it's own set of delays.

Is there a location where the latest version can be pulled/tracked with Obtainium?

 

Almost 30 more minutes of dishwasher.

 

This may violate the "low-effort" rule, but I thought readers might enjoy (remove if not, of course). This particular Android app Privacy Policy shows how the "one size fits all" of the Play Store doesn't always make sense.

 

cross-posted from: https://lemmy.nowsci.com/post/7576006

Hey all,

In the market for a GPU, would like to use Bazzite, mostly a Steam user with some SteamVR (rare), and have run into nvidia issues with Bazzite and a 3070 previously.

With the recent news on nVidia's beta drivers and Plasma's sync support in beta, I'm newly on the fence about switching to AMD given nvidia having a better performance to cost ratio, the power usage (big one for a compact living room system), and the fact that they have the potential for HDMI2.1 support which AMD doesn't have a solution to yet.

What are community thoughts? I'll probably hold out for some reports on the new drivers regardless, but wanted to check in with the hive mind.

Thanks!

 

Hey all,

In the market for a GPU, would like to use Bazzite, mostly a Steam user with some SteamVR (rare), and have run into nvidia issues with Bazzite and a 3070 previously.

With the recent news on nVidia's beta drivers and Plasma's sync support in beta, I'm newly on the fence about switching to AMD given nvidia having a better performance to cost ratio, the power usage (big one for a compact living room system), and the fact that they have the potential for HDMI2.1 support which AMD doesn't have a solution to yet.

What are community thoughts? I'll probably hold out for some reports on the new drivers regardless, but wanted to check in with the hive mind.

Thanks!

258
This is real rule. (lemmy.nowsci.com)
submitted 1 year ago* (last edited 1 year ago) by fmstrat@lemmy.nowsci.com to c/196@lemmy.blahaj.zone
13
submitted 1 year ago* (last edited 1 year ago) by fmstrat@lemmy.nowsci.com to c/lemmy_support@lemmy.ml
 

Hi all,

Hoping someone can help me out here. I just did the upgrade, I've updated my ENV variables, and started, and I get:

2024-05-18T19:13:07.047224829Z 2024-05-18T19:13:07.046998Z  INFO migrate_04: pict_rs::repo::migrate: Running checks
2024-05-18T19:13:08.177484313Z 2024-05-18T19:13:08.177294Z  INFO migrate_04: pict_rs::repo::migrate: Checks complete, migrating repo
2024-05-18T19:13:08.177524752Z 2024-05-18T19:13:08.177323Z  INFO migrate_04: pict_rs::repo::migrate: 362224 hashes will be migrated

I then proceed to get the following item for 110,670 files:

2024-05-18T19:13:08.196186995Z 2024-05-18T19:13:08.196036Z ERROR pict_rs::repo::migrate: Failed to migrate hash 0000181884f6a92c7f6ee411dc8049474b2872204fafebbc6b0170e7499d902b, skipping
2024-05-18T19:13:08.207066220Z 
2024-05-18T19:13:08.207078792Z    0: Error in store
2024-05-18T19:13:08.207089618Z    1: Requested file is not found
2024-05-18T19:13:08.207099605Z    2: No such file or directory (os error 2)
2024-05-18T19:13:08.207109313Z 
2024-05-18T19:13:08.207119231Z Location:
2024-05-18T19:13:08.207129638Z    /drone/src/src/repo/migrate.rs:32

And then pict-rs just exits.

Turning on debug gives me more info, but nothing helpful after each of the above errors:

|lemmy-pictrs  | 2024-05-18T19:21:15.631508936Z 2024-05-18T19:21:15.631417Z DEBUG sled::pagecache::iobuf: advancing offset within the current segment from 102576771 to 102576797    
|lemmy-pictrs  | 2024-05-18T19:21:15.631606646Z 2024-05-18T19:21:15.631504Z DEBUG sled::pagecache::iobuf: advancing offset within the current segment from 102576797 to 102576823    
|lemmy-pictrs  | 2024-05-18T19:21:15.631643803Z 2024-05-18T19:21:15.631531Z DEBUG sled::pagecache::iobuf: wrote lsns 3029152387-3029152412 to disk at offsets 102576771-102576796, maxed false complete_len 26    
|lemmy-pictrs  | 2024-05-18T19:21:15.631661473Z 2024-05-18T19:21:15.631541Z DEBUG sled::pagecache::iobuf: advancing offset within the current segment from 102576823 to 102576849    
|lemmy-pictrs  | 2024-05-18T19:21:15.631676419Z 2024-05-18T19:21:15.631549Z DEBUG sled::pagecache::iobuf: mark_interval(3029152387, 26)    
|lemmy-pictrs  | 2024-05-18T19:21:15.631689340Z 2024-05-18T19:21:15.631561Z DEBUG sled::pagecache::iobuf: new highest interval: 3029152387 - 3029152412

My env (left old values in, too):

      - PICTRS_OPENTELEMETRY_URL=http://otel:4137
      - PICTRS__API_KEY=...
      - PICTRS__SERVER__API_KEY=...
      - RUST_LOG=debug
      #- RUST_BACKTRACE=full
      - PICTRS__MEDIA__VIDEO_CODEC=vp9
      - PICTRS__MEDIA__GIF__MAX_WIDTH=256
      - PICTRS__MEDIA__GIF__MAX_HEIGHT=256
      - PICTRS__MEDIA__GIF__MAX_AREA=65536
      - PICTRS__MEDIA__GIF__MAX_FRAME_COUNT=400
      - PICTRS__MEDIA__ANIMATION__MAX_WIDTH=256
      - PICTRS__MEDIA__ANIMATION__MAX_HEIGHT=256
      - PICTRS__MEDIA__ANIMATION__MAX_AREA=65536
      - PICTRS__MEDIA__ANIMATION__MAX_FRAME_COUNT=400
      - PICTRS__MEDIA__VIDEO__ENABLE=True
      - PICTRS__MEDIA__VIDEO__MAX_FILE_SIZE=20

Any idea how to get pict-rs started? Or what I should look to for figuring this out.

Edit: Scanning the logs more closely I see that I never get the Migration complete message from here https://git.asonix.dog/asonix/pict-rs/src/commit/d45e3fa386e62e0538b3d7ba399ceaec80fdd7ca/src/repo/migrate.rs#L175, but I do get percentage completes up to Migration 59% complete - 213698/362224. Nothing else indicates, though.

view more: ‹ prev next ›