BuoyantCitrus

I've blithely assumed that backups / snapshots of my home dir (including my Thunderbird profile) were covering my email. But it occurs to me it may be more difficult than expected.

I have message synchronization on for any folders I care about ("for offline use"). What I was assuming this meant was that if my mail host disappeared or mysteriously deleted an important folder, I would still be able to switch to a backup, start TB in offline mode (via a commandline parameter), and copy those folders to a local folder at which point I could reconnect and drag them back to my new host, a local imapd I use as an archive, or wherever.

But ...would that actually work? Anyone recover email from offline folders? How'd that go?

Edit:

Well, there can never be too many reminders to verify our backups and I'm all for that but that's less what I was after. I was specifically thinking about the scenario when an IMAP host somehow loses an important folder or disappears entirely. How would it go to recover from a sync'd folder in tb? What caveats would there be? Would attachments show up?

But ya, this post was silly, it's easy enough to try. Yes it works, yes the attachments come with. No major issues in my limited test.

However, I did learn one annoying thing: there is no command line option to start Thunderbird in offline mode. So in the case where the folder was deleted on IMAP I'd either have to:

• disconnect from the network before running the app
• quickly toggle offline before it finishes connecting and deleting the folder
• use the pref to prompt if you want to go online every time you start

I think for as rare a scenario as this is it's fine to just disconnect but I'm a bit surprised it really doesn't seem to have a flag for it.

I see there's an update coming soon that will add support for AVIF (woo!) and I wonder if that'll also coincide with enabling WebP for pixelfed.social? I was hoping to use less platform resources by uploading smaller/better files.

Also, if they're smaller maybe they won't have to be reconverted server-side? It'd be nice if I could optimize them locally from RAW without them being reprocessed but didn't see any guidelines in help that would guide me in doing that. Or will it be re-jpg'd regardless of what I send?

I know it's my fault for believing what my neglected laptop told me about its battery but I went ahead an did a kernel update anyway and wound up needing to repair my system.

After a quick search I wound up on https://wiki.debian.org/GrubEFIReinstallOnLUKS per usual.

The biggest hassle of this is having to type out the longish for loop to bind the various vfs to the chroot environment. It was bad enough when it was proc/sys/dev but it's worse these days:

for i in /dev /dev/pts /proc /sys /sys/firmware/efi/efivars /run; do sudo mount -B $i /mnt$i; done

I realise there are various things that'd automate that if I connected the rescue image to the internet and added a package but that's also hassles as I've really just booted it with the express purpose of reinstalling grub.

But maybe there is already some form of shortcut for this in the system that I've missed? Or some existing ticket/effort to enact one I could +1?

My Keychron Q11 showed up recently and I've been super happy with it. Main reason was that my Noppoo Choc Mini finally lost a switch and I don't have any on hand (nor a soldering iron ...yet) but it turns out I actually really wanted the pair of rotary encoders on this and didn't even realise.

Specifically, I've got it bound to Ctrl-PgUp/PgDown so I can scroll through my tabs with it and close them with a click binding to Ctrl-W and that's working out really well.

Anyone else use the knobs like that? I've got the other one set to volume and the vendor had zoom as a suggestion but I wonder what else people do with these?

Bonus newb Q: On the product page they demonstrate binding Ctrl-+ zooming to the encoder via a macro but neither macro13 nor the {KC_LCTL,KC-W} type syntax would let me click "Confirm" when trying to associate it to the knob in Via (eg. it wouldn't let me follow their example). Luckily it was happy with the alternative of LCTL(KC_W) that I stumbled on somewhere but now I wonder how to properly associate a macro to a knob?

Last time I needed to add rf to a desktop, Intel AX200 seemed like the chipset to get. But now there are various new standards and the BE200 apparently has issues with AMD systems? So is there something newish from Qualcomm or others that I should be aiming for or would I probably be better off just picking up an AX210?

Since the card might be kicking around a while I'm curious what has the best overall Linux support with as many significant 802.11 standards and Bluetooth codecs as possible for general future-proof-ness. Would also be nice if it had good support for AP mode as that's sometimes handy or I might repurpose it into a router at some point.

Apparently, while it's closed for new donations, liberapay is still going to renew existing ones.

cross-posted from: https://lemmy.ca/post/1926125

Too many perfectly usable phones are put into a questionable security situation by lack of vendor support for keeping key software up to date.

But what's the actual risk of using an Android phone on a stock ROM without updates? What's the attack surface?

It seems like most things that'd contact potentially malicious software are web and messaging software, but that's all done by apps which continue to receive updates (at least until the android version is entirely unsupported) eg. Webview, Firefox, Signal, etc.

So are the main avenues for attack then sketchy apps and wifi points? If one is careful to use a minimal set of widely scrutinised apps and avoid connecting to wifi/bluetooth/etc. devices of questionable provenance is it really taking that much of a risk to continue using a device past EOL?

Or do browsers rely on system libraries that have plausible attack vectors? Perhaps images, video, font etc. rendering could be compromised? At this point though, that stack must be quite hardened and mature, it'd be major news for libjpg/ffmpeg to have a code-execution vulnerability? Plus it seems unlikely that they wouldn't just include this in webview/Firefox as there must surely be millions of devices in this situation so why not take the easy step of distributing a bit more in the APK?

I'm not at all an Android developer though, perhaps this is very naive and I'm missing something major?

Too many perfectly usable phones are put into a questionable security situation by lack of vendor support for keeping key software up to date.

But what's the actual risk of using an Android phone on a stock ROM without updates? What's the attack surface?

It seems like most things that'd contact potentially malicious software are web and messaging software, but that's all done by apps which continue to receive updates (at least until the android version is entirely unsupported) eg. Webview, Firefox, Signal, etc.

So are the main avenues for attack then sketchy apps and wifi points? If one is careful to use a minimal set of widely scrutinised apps and avoid connecting to wifi/bluetooth/etc. devices of questionable provenance is it really taking that much of a risk to continue using a device past EOL?

Or do browsers rely on system libraries that have plausible attack vectors? Perhaps images, video, font etc. rendering could be compromised? At this point though, that stack must be quite hardened and mature, it'd be major news for libjpg/ffmpeg to have a code-execution vulnerability? Plus it seems unlikely that they wouldn't just include this in webview/Firefox as there must surely be millions of devices in this situation so why not take the easy step of distributing a bit more in the APK?

I'm not at all an Android developer though, perhaps this is very naive and I'm missing something major?

Allied Properties sale of their data centre portfolio to KDDI includes 151 Front Street W., the site of TorIX which is the main Internet Exchange Point for the country. While that's not necessarily an issue, I kinda figured it was at least a little bit notable but I've not seen it mentioned aside from an investment context.

Unfortunately, it seems like it's less consequential than it should be because Bell Canada apparently still refuses to peer at TorIX and only connects to other ISPs through the US which means that eg. if I'm on Rogers in Toronto and you're on Bell, any communications between our computers have to flow through American controlled systems even though we're in the same city because that's how Bell chooses to have things set up.

Whereas, for pretty much everything else in Toronto, it'd move between networks via TorIX. Which is now in a building owned by a Japanese company instead of a Canadian REIT.

Was curious about whether someone could extract my password from Jerboa on my phone but didn't get any response there. Maybe you guys have some idea? Does Lemmy even offer an auth mechanism that could prevent this, is one in the works?

cross-posted from: https://lemmy.ca/post/652328

I noticed that Jeroba didn't seem to switch to a different site the way Relay passed through to Reddit so I could log in and link it via OAuth. From that I take it that when I authenticate in Jeroba I'm entrusting it with the cleartext password for my lemmy account which it's storing on my phone?

I'm sorta okay with that especially for now (eg. alpha) so I proceeded with things but maybe it should be more clear up front that's what's happening? And really, any of the other apps could probably have faked that OAuth page anyhow so it's dubious if you were really trusting the app all that much less in that case.

However, one thing OAuth had going for it was that would make it a lot harder for someone who steals my phone to permanently take control of my Reddit account whereas they could extract my password from Jeroba and use it to take over my lemmy account?

Looked through the docs a bit and it's not really clear to me: I'm posting this on lemmy.ca, does that mean only that instance knows my IP? Or does every instance it federates with get my ip alongside this post?

This seems maybe important, did I miss a privacy guide to Lemmy someplace? Cursory searching didn't come up with much official. Are there other aspects we should be thinking about here? I'd come across some mention of deleted posts being still available everywhere they were sent but that sorta makes sense -- hard to "unpublish" anything.