• 10 Posts
  • 28 Comments
Joined 9 months ago
cake
Cake day: October 11th, 2023

help-circle
  • According to my recent test, the premium was 4.7 percent compared to spot rates, hence their pricing is not competitive and Bitrefill remains without a serious contestant. There you’d only pay the ~0.5% fee for going through an instant exchange in order to have your XMR arrive as BTC.

    EDIT: Now, a day later, I did another test and got percentages from 1.8 - 2.0 % which is much more reasonable.

    Hint: To quickly get the hidden fees of any purchase, execute units like this: ./units.sh '<xmr-cost-at-checkout> XMR' '<EUR|USD|...>' or ./units.sh '<xmr-cost-at-checkout> XMR / <value-in-fiat> <EUR|USD|...>' '%' for the total percentage asked.








  • more aliases are available to register

    This one is technically not true until you add Punycode support - and only if you manage to remain below XMR.ID’s user count by that time :D

    (Without Punycode, staying RFC-compliant, and applying XMR.IST’s restriction of 30-characters max, we could provide roughly a count of 30^37-1-<amount of users>, but even if we had a 10-chars limit, the number would still be unfathomable.)

    Welcome to the space - it feels less lonely now!















  • The time required depends on where you start. Someone who knows how to register a domain but has yet to read up on OpenAlias will probably need about an hour or two (if we do not take into account the hassles associated with DNSSEC with many registrars).

    Then the cost of a privately registered domain starts at around 15 dollars per year, whereas the same is roughly the one-time price of a permanent XMR ID with two domains secured against each other (meaning that both, DUKETHORION.xmr.id and DUKETHORION.xmrid.com will return the same Monero destination). Wallets can opt to verify this.


  • Manipulation of any record would immediately trigger a notification to all affected users, leaving me with nothing but a destroyed reputation.

    The most granular use I can think of is telling someone in-person to load your XMR ID on their device and then confirming what you see.

    Coupled with a client that stores the result in a local address book - and compares it with the current DNS responses every time - even senders can be sure that they are still working with valid information.

    (An extension to the official Monero client supporting this is in the works.)




  • Whenever someone would publish their experiences with AllArk on Reddit, the thread would get “downvoted into oblivion”, with some people raising concerns of them themselves using bots to do so.

    I also remember seeing an email log of a user “losing” a couple of hundreds to that same entity. I thought that was on monero.observer, but cannot seem to find it now.

    Maybe someone else feels inclined to dig something up, but in general, just be very careful with what people recommend in the space. Most of the time they saw it mentioned somewhere and, with best intentions, just pass it on.


  • This is bad news for those of us who were not only looking to give old mobile hardware a longer lifespan, but simultaneously obtain privacy and security while doing so.

    The arguments provided in the blog post are rather faint and give a vibe of “holding on to last straws”, as other distributions and even BSD’s have managed to run both GNOME and KDE fine, even before pmOS.

    For readers unfamiliar with systemd’s drawbacks, these resources can serve as good starting points:

    without-systemd.org // nosystemd.org


    Out of curiosity: Can you point to a log of the communication with the Alpine team?