Messages in this thread Patch in this message | | | From | (david parsons) | Subject | Arguing about backwards compatability (+ a tiny fcntl locking patch) | Date | 23 Jun 1999 11:47:27 -0700 |
| |
In article <linux.kernel.Pine.GSO.3.96.990623125723.9662K-100000@joni.pasture.net>, Robert Minichino <linuxkrn@denarius.com> wrote: >On 22 Jun 1999, david parsons wrote: >> david parsons \bi/ At 100k code-bloat per major release, there are many worse >> \/ things to worry about than whether it's time to kill >> off the stable interfaces > >But in 15 years, those interfaces -will- still be around if they're not >yanked out. Also, much of the "code-bloat" in the major releases is the >addition of drivers and other code that can be conditionally compiled.
No it isn't. I've kept pretty much the same set of drivers in my standard kernel since 1.2.13 days, and the kernel size went from about 400k (1.2.13) to 500k (2.0.28+logo) to 600k (2.2.10) to 640k (2.3.6) Some of this is additional driver (VIA IDE support, but those drivers came into existance at the same time AMD Lance support was modularised, so it's a wash)
>With a completely open development cycle, and massive amounts of >communication directly to the developers available (such as with >linux-kernel), any changes made to internal kernel interfaces (which are >seldom unnecessary) are open to discussion to any interested party.
Linux isn't a democracy. I can discuss until my face turns red, but if backwards compatability isn't important to any kernel developer (I am not a kernel developer; I just hack the kernel so that it works with my projects) it won't make any difference.
As a general rule, it's a Good Thing that the people who write the code get to make the rules, but it's also NOT a Good Thing to assume that everyone is a hobbyist who is willing to patch their codebase every time a kernel interface is tweaked.
And speaking of patches, this patch to 2.2.10 puts back the old-style fcntl locking:
[cut here] --- linux-2.2.10-orig/fs/locks.c Tue May 11 08:52:14 1999 +++ linux-2.2.10/fs/locks.c Wed Jun 23 11:47:11 1999 @@ -441,10 +441,10 @@ current->pid, current->comm); } } +#endif if (!(filp->f_mode & 3)) goto out_putf; break; -#endif default: error = -EINVAL; goto out_putf; [cut here]
____ david parsons \bi/ This patch also works with 2.3.6. \/ - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |