Whats up with the constant software updates?

Lets say i have a woking system from 2018. Why would i want to update stuff while having a working system, just to get stuff potentially broken with updates or have mandatory updates to related software?
Is there a need for new stuff every so
often? If it works dont "fix" it.
A well designed program imo works with the minimum, most standard possible api calls and fewer libraries possible. I think there is an overengineering problem in software. Sure a developer can experiment as long as he likes but on test systems. And i suspect there are big players (software and hardware vendors) that deliberately break interfaces all the time and drag the rest of the industry with them by perceived necessity.
On the other hand admittedly, robust usable problems have sometimes the opposite problem, they dont have people to work on them.
 
If someone does not care about security, then there is no need to update a system. But then if the system is compromised because of an unpatched security hole, then they can't blame anyone but themself.

No one in FreeBSD has ever forced a user to update. Updates are generated for reasons, but unlike MicroSoft/Apple/Android, the user must actively apply the update.
 
The setup and configurations become stale over time as other software changes. Routine updates act as a form of re-sync with the newer versions minimizing the impact of a big jump between a release from 2018 and today.

Periodic syncs also keep the understanding of the system up to date. One does not need to swap tools all the time, keep using whatever works, but IMHO do update the system.

The longer the system lives, the larger the gap becomes, the higher the chances are that something will break in the future.
 
Everything has bugs (like that 15yr old file copy one fixed with newest KDE), and anything can be an improvement (Jay + GPL shaders on non-Xe); I update everything until there's a reason not :p (like Bun >1.13.4)
 
[Mod hat: Probably got double posted due to some lag we've been experiencing. As both had responses I merged both threads instead of deleting the double..]
 
I think OP hints at a larger issue that bugs me about the industry: continuous integration. It is abhorrent in many ways. It trivializes software development to being an assembly line process, it fosters the lie of "software is a service" instead of a product. It is tightly coupled to "agile-crap", etc. The only models I've appreciated were a hybrid approach that merges code-to-specs and test driven with iterative development where goals are clearly outlined and must be signed off on by the stakeholders at every step of the way. In my world this leads to fixed release schedules and discourages continuous integration, except where critical bugs are found...but then, following my approach leads to much fewer bugs anyway.
 
Sometimes fixing individual bugs requires a rework of existing things. That rework might introduce new bugs.

Even without adding features there is constant bug churn.
Yes, but for some reason new software tends to break compatibilty with older libraries too easily, if it was just bug fixes things would mostly work out of the box, except maybe the few programs that were programmed based in other softwares bugs/glitches. Maybe its just a habit for some developers to put newer compile time dependencies while their code isnt actually using them? I dont know. GPU driver stuff is especially annoying too in my small experience.
 
You
I think OP hints at a larger issue that bugs me about the industry: continuous integration. It is abhorrent in many ways. It trivializes software development to being an assembly line process, it fosters the lie of "software is a service" instead of a product. It is tightly coupled to "agile-crap", etc. The only models I've appreciated were a hybrid approach that merges code-to-specs and test driven with iterative development where goals are clearly outlined and must be signed off on by the stakeholders at every step of the way. In my world this leads to fixed release schedules and discourages continuous integration, except where critical bugs are found...but then, following my approach leads to much fewer bugs anyway.
Yeah we live in an era that fights against humans decision potential. The programmer gets delegated from a project general to a camp janitor. Maybe a few exceptions here and there.
 
Security.
Everybody's talking about bugs but this is just as valid a point. Some security hole gets discovered that's been sitting there for years and someone found it. So the hole has to be fixed but then it's discovered something else breaks cause that hole was depended on. It was never a problem before. Or fixing the security issue unveiled other security issues later on.

If there were no problems then there would be no need to upgrade. So it's either security or feature adds. Which is why it's silly when someone says some software can't be any good cause it's old and hasn't been touched in a long time to which I always reply, "Maybe it's done!" Age is never any indication of the quality of software.
 
Back
Top