• 0 Posts
  • 21 Comments
Joined 1 year ago
cake
Cake day: June 1st, 2023

help-circle
rss
  • As I understand it, these kind of applications depend on being able to perform activities in the background, which is highly limited in iOS for battery efficiency reasons–and maybe for privacy.

    Many years ago I was working on a project that shared connectivity details over wifi/bt, and iOS was troublesome also due to the application not being aware of the local bluetooth address.

    Possibly similar issues impact other mesh networking applications on the platform.





  • Too bad, would’ve considered it as a viable option to mdadm + BTRFS.

    Currently I’m using bcachefs with LVM (which can do raid, but I currently only have one NVME SSD), though it indeed does have RAID1/0/10 support. But overall I expect it not to not make the same silly default choices as btrfs, such as not being able to start the system if a RAID1 component of your root filesystem is missing. And, supposedly, when the RAID5/6 becomes stable, it won’t have the write hole problem.

    It said the code base was build on something stable, but it didn’t say what, do you happen to know what FS this project is a fork of?

    It’s based on bcache :) by the same author, but of course bcache is not really a file system but rather some kind of object storage layer for the purpose of caching slower block devices and absorbing write load.

    Bcachefs might be coming soon to the mainline kernel, so that’s going to make it a lot easier to try out. Personally however I have lost one bcachefs (that FS was readable, though, and I have good backups), but I have also lost a btrfs before and seen reiserfs bugs, so I don’t too heavily count it against it; overall I enjoy its stability when using basic functionality. I haven’t dared trying snapshots with it yet…


  • Depends on how much you change per time unit.

    I take full system backups every three hours, but the backups are thinned so that there are previous 24 hourly ones, previous n daily ones, previous m monthly ones, etc. Similar approach can be used with snapshots.

    I don’t currently use snapshots—I don’t run btrfs anymore—but when I did, I did a snapshot every hour and kept them for 24 hours. But then I backed up the latest snapshot, which gives consistent backups, versus regular backups where files can change while you’re doing them. I’m nowadays using bcachefs, but I don’t quite trust its snapshots yet so I haven’t started using them ;).


  • Does it though? https://www.europarl.europa.eu/doceo/document/TA-9-2023-0237_EN.html#title1:~:text=(39)   To,by end-users. does say

    To ensure the safety of end-users, this Regulation should provide for a limited derogation for portable batteries from the removability and replaceability requirements set for portable batteries concerning appliances that incorporate portable batteries and that are specifically designed to be used, for the majority of the active service of the appliance, in an environment that is regularly subject to splashing water, water streams or water immersion and that are intended to be washable or rinseable. This derogation should only apply when it is not possible, by way of redesign of the appliance, to ensure the safety of the end-user and the safe continued use of the appliance after the end-user has correctly followed the instructions to remove and replace the battery. Where the derogation applies, the product should be designed in such a way as to make the battery removable and replaceable only by independent professionals, and not by end-users.

    (emphasis mine)

    I dont think it would be acceptable to argue a regular consumer phone would fall into that exception.



  • I believe you’re completely right here, except that snapd can be configured to point to another store, though it’s not very well documented… I did find the piece of information once :).

    But the thing is that the client still only supports one app backing site at a time. So if you pick another one, you lose visibility to the other store. I doubt even updates work as they should.

    So it’s really about building technology that is geared towards centralized control, whereas basically anyone can host flatpak packages and give ref links to them.



  • I don’t think publically available blockers can really win this battle in the end. After all, in the end game Google could just

    1. setup a system that runs a browser
    2. downloads the updates as they come
    3. automatically modify the system so that blockers are detected or they fail to block it

    This is possibly even relatively easy with the help of LLMs nowdays.

    On the other hand, Google backend code is completely secret and for frontend and protocols they can apply opfuscation techniques, requiring manual updates by blocklist maintainers or adblock developers, taking a lot of time continuously. I suppose LLMs could help here as well, but it’s harder and such attempts could even be detected by Google, because they would need to be tested against their system.


    The only solutions I see are to move on from Youtube, have private blockers that don’t become too popular, or tolerate the ads.


  • I don’t believe keeping people hooked up was the driving force when e.g. upvotes or subscriptions were designed and implemented for e.g. Lemmy. They are just quite practical and nice features to have.

    I suppose if the argument is (is it?) that making sites nicer to use makes them also addictive—and I suppose in some sense that is true—then my counter-argument would be that it doesn’t mean that the reason for making them nicer to use was to induce addiction.





  • nc is nice an simple tool that’s been useful for quite a bit of times in the past.

    However, perhaps take a look at socat, which is a lot more versatile tool for connecting two kinds of file descriptors, that may refer to files, standard input, standard input with readline, TCP/UDP/SCTP/UNIX/multicast, supports proxies—and also handles end-of-stream properly that I at least have had trouble with nc, but maybe it’s fixed nowadays.

    Basic socat invocation: socat - tcp4:1.2.3.4:22. Can also do tcp4-listen:1234 and indeed by combining those you get a TCP forwarder. My most common use case is as an ssh ProxyCommand, socat - tcp4:%u:%p. (Well not that directly, but effectively doing that from within a shell script that decides where to connect, to handle a changing IP.)

    socat doesn’t come with port scanning though, but perhaps better use specialized tools, such as strobe or nmap for that use anyway.

    There’s also a modern alternative to socat: websocat. It’s name-sake trick is to work with websockets, but seems to also do most of what socat can plus it can also do filtering with “overlays”. I haven’t really used that tool, though, but maybe I will if I need to deal with websockets from command line.