On 3/24/21 6:00 AM, Volker Hilsheimer wrote:
Who said anything about "ever?" Or even "last year." Have you never built
anything that is actually finished, and stays finished, and relevant, and functional for 10, 20, 30
years? Granted, if I get 10 years of use out of anything built in this century, I do consider it a
minor victory. So maybe that's my answer.
There are evidently (form what has been written here, and from my personal
experience of working in and with financial, medical, and telko) industries
that prefer 30 year old security issues in their devices over establishing a
process that allows them to continuously update their software stack. In which
case, yes I personally do think they are missing the point of “software”, and
I’m happy that our way of developing Qt is not constrained by those industries.
Those would be your customers. 30 year old security issues tend to be
easily plugged or otherwise defended against. There is no defense
against the relentless stream of new security vulnerabilities, bugs, and
crashes brought about by continuous integration and deployment.
The Wild Wild West coding style really does kill people. A blue screen
of death on your heart surgery robot mid-surgery generally results in a
real death.
Yes it’s great if stuff that some of the stuff I built 30 years ago still works
(at least I assume it does; I’d have to find a working floppy drive, and a DOS
emulator with Turbo Pascal). But then I don’t expect it to look and feel great
on the latest macOS version.
I don't have to. The Parts Order Routing System at Navistar has been
making sure trucks have the parts and maintenance items they need to
keep bringing food to your store, supplies to your construction sites
and service vehicles for when your power lines go down. It is also
keeping the fire engines and pump trucks in many/most places running,
not to mention keeping the Blue Bird buses safe and operational hauling
your kids to school.
The trading floor system at the Chicago Stock Exchange I first worked on
in the late 1980s ran to 2005/6 when they got rid of the trading floor
to become an all electric stock exchange.
To be honest, many of those bugs are really hard to fix without breaking
anything else, so often we decide that a known, well-documented bug is
preferable to a bunch of new, unknown bugs that a fix might introduce.
You knowingly create 30 year old security issues and you diss your
customers who have worked around theirs.
:P
I'd go on to explain that not everyone who uses Qt, lives and breathes Qt,
finds it endlessly fascinating, nor has time to conform to your strict and
relentless release schedule to evaluate every nuance of massive API changes,
etc, etc.
Apropos generalising. Are we still talking about the first release in 7 years
where APIs were changed?
FWIW, so far the substance of this discussion seems to boil down to
* the old QList implementation being gone
* toContainer convenience methods removed
* references to QHash entries no longer stable when the hash is mutated
You forgot customer abandonment
death of OpenSource LTS
Qt 6 being useless
QML needing to be ripped out.
--
Roland Hughes, President
Logikal Solutions
(630)-205-1593
http://www.theminimumyouneedtoknow.com
http://www.infiniteexposure.net
http://www.johnsmith-book.com
http://www.logikalblog.com
http://www.interestingauthors.com/blog
_______________________________________________
Interest mailing list
Interest@qt-project.org
https://lists.qt-project.org/listinfo/interest
