On Fri, 2023-02-17 at 16:51 -0800, ToddAndMargo via users wrote:
> It is the better part of wisdom to realize when
> computers are not the proper tool for the job.
>
> I hope this catches up to the cloud soon. Somethings
> are great on the cloud; others are better on the
> edge. And I am seeing a lot of the "are not" going
> to the cloud anyway.
[...]
I never got into modern GUI system programming. The notion of having
to frankenstein your program from a plethora of other people's
libraries, with unknown problems, unfixable problems, inadequate
information, and an ever-changing landscape of libraries and OS was
more than I was willing to get involved in.
Yes. A lot of software follows the same practices as the automobile industry --
adding tailfins and other "features" to keep people buying new versions when
the old version did the job. Many people seem to enjoy twiddling with new
interfaces to figure how they work when the older interface provided nearly
all the necessary functionality (and both old and new interfaces neglecting
needs of users with disabilities).
Software is all that and worse due to the ongoing lack of remedies for reliance by
Todays case in point:
There are useful innovations in all fields, but most people can't sort of the useful
from the useless, and the best innovations are often made by small groups that
don't have name recognition.
Sandy Munro has a good insights, based on years working in the industry, into these
issues for the automobile industry -- we need people who can do the same for software.
One of Sandy's current complaints is the auto EV makers who use an array of 3rd
party controller models, each requiring a case, power, and data cables, instead of
putting all the functions in one module. All those added cables and connectors are
potential failure points and added cost for every vehicle sold.
In the linux world, with X11 provided users could work in a terminal and start the GUI
from the command line, but over the years, features were added using libraries of
varying quality and complicating the interfaces. Wayland et al addressed many of the
shortcomings, but was tied to efforts that appear to be aimed at providing a GUI that
would be familiar to users whose only experience with computers was with smartphones.
My field (ocean remote sensing) revolves around geolocated images. The NetCDF4-CF
file format was developed with inputs from diverse communities, and is used for much of
data I encounter. Many users, however, rely on internet searches to "learn" that
they need geoTIFF files. Unfortunately, geoTIFF lacks the comprehensive metadata
support you can get using NetCDF4-CF files. When I provide files to users who
insist on geoTIFF files, I need to translate the NetCDF4-CF metadata into text documents,
and the end user has to figure out how to enter the information into their GIS package.
I find I'm getting sucked into using Windows, macOS, and Wayland on linux because that
is what colleagues in large enterprises are using. Certainly in my former work, IT was
build around a call-center model with no regard for the impacts on actual workflows.
--