An (incomprehensive) list of missing features in technology. Last update: 2022-06-01
missing features in/of computer technology
- Easy to setup and use, distributed, cross plattform, file synchronization. Should be implemented as a file system feature, using the techniques developed for and used by distributed version control. Think Git, but as a ZFS feature with
zfs send /
zfs recv instead of Git push/pull.
- Approachable tiny scale database systems that enable and make laypersons to create proper tools for their hobby / club / stamp collection / etc. instead of doing spreadsheet "databases".
- Content aware filesystems. Several attempts over the years, none of which worked out. Seems to be a hard problem, difficult to reconcile tags/semantics with the strict heirarchy of path based file orgnaization.
- Layman friendly installation of a complete operating system + application suite, including configuration of a well laid out communication and productivity tool setup, all in a fully encrypted and measured boot protected environment, which does not present a walled garden. Apple delivers a coherent experience; yet under the hood it's a complete and utter mess, and of course it's a walled garden with razor wire on top. Windows-10… user hostility as a service; sure it can do secure boot, but protect itself from the user, not other way round. Linux and the *BSDs are neither user friendly, nor easy to setup, Ubuntu is user hostile.
- Layman friendly installation of self hosted essential services; federated (ActivityPub) web presence, email, messaging. VPS are dirt cheap these days and unlike you're running an online business, bandwidth and latency are no huge issues, even if hosting video. You're reading this from such a dirt cheap VPS right now. Heck, this opens even further business opportunities for CDN providers to allow for easy connection with such self hosted, personal internet presences.
- UIs quickly and consistently responding to theme/style changes. This did actually regress over the past 1½ (give or take) decades:
- On Win32 applications get immediate notification via
- On X11 technically we have the Xresources properties on the root window. Applications can select on changes of that property and react to it. "Old school" applications actually do. But GNOME/GTK+ and Qt developers either are to incompetent, unknowledgable or just don't give a fuck to use that, and instead reinvent that wheel – poorly – by means of at first dotfiles and later on a NIH replacement of Xresources called XSettings, which rely on a settings daemon to hold the key value store and actively ping all interested clients of the change in the name of "efficiency". Did those morons actually profile that thing and compare the overhead?
- Wayland? Oh you mean GTK+ and Qt and however those two get along with each other? Mostly see X11, but since at the moment Wayland implies GTK+ and/or Qt for anything other than games and Blender I'll just pack it here: Who in their right mind was thinking that CSS was a viable method for styling and layouting toolkit widgets? CSS already is terrible on the web. Why shove it down our developer throats outside the web as well?
- macOS: "we don't need no stinking theming. Here, you can change the color of the selection highlight, now fuck off!". Of course some applications that rely on a different color contrast to support the work done with them (color grading, image manipulation) will just happily ship their own toolkit, programmed to look like the stuff macOS offers, which – thanks to the lack of theming and user defined coloring – is easy enough.
- A free access for all infrastructure for bootstrapping, general purpose, application agnostic distributed hash tables. Yes, you can piggyback on the BitTorrent bootstrap servers. But I don't think it's nice nor long term stable. I'd like to see something like that implemented alongside the DNS root server. Heck, you could build the whole thing on top of DNS inside a new
- "Ad-Hoc" content streaming fanout being part of fundamental IP routing; i.e. some kind of multicast content distribution, which does not rely on infrastructure controlled by a single entity. Existing multicast relies on allocation of a global multicast address and as far as implementations go, it's more miss than hit.