I said that they are a layer away - apps go through the system shared libraries to access kernel functions in most cases, so reallynare well isolated. 36 years in *nix as a professional, including System-V at the source level, taught by AT&T, (If Win*/MS does something tupid, I'm not at all surpised . . . ). Essentiallynhave my own Linux distro, and can't recall a libc update forcing a kernel update (or the opposite) other than the change away from a.out binary format. I do load software from my own source builds wherever possible though, due to app symbol/lib version issues . . .).
переключити цитоване повідомлення
Показати цитований текст
On December 29, 2024 8:16:02 PM CST, Jerry Stuckle <ai0k@...> wrote:
Tim,
I've worked on multiple APIs in the 50+ years I've been programming (including as an IBM employee) on everything from MS-DOS 1.0 to mainframe operating systems like MVS.
Yes, few applications interface directly with the kernel. But the API consists of two parts - the interface and the implementation.
The interface is what the applications see. The implementation is how it interacts with the kernel and other resources. A change in the kernel can definitely change the implementation (which is why you have version-specific libraries).
But sometimes the changes in the kernel can't be covered by the implementation without changes to the interface. And yes, it does happen which is why application sometimes have version specific libraries.
So you can't say that changes in the kernel do not affect applications. They definitely can. That may or may not be the problem here but not knowing the reason for the rejection the biggest thing I see between being able to download or not is the kernel version.
Jerry, AI0K
--
Sent from my Android device with K-9 Mail. Please excuse my brevity.
