On Sun, Sep 22, 2002 at 02:23:20AM +0300, Mark Veltzer wrote: > It's better than 2.4.18 redhat but actually you're always better off using > vanilla (the ugly driver code that RH puts in which rightly doesn't belong in > the kernel isn't something you want even if you have the specific hardware). > I wouldn't get an -pre from Marcello. If I would go for non vanilla I would > take stuff off the -ac tree (Alan is finding lots of bugs and fixing stuff in > the IDE layer lately and I like that...). If you're not a kernel hacker I
Alan's tree is really exeperimental right now, due to the IDE work going on in it. You should only use it if you're willing to suffer the consequences... > don't see any reason what so ever to use a non vanilla (you're just wasting > your time unless your objective is to experiment with kernels in which case > go right ahead and get your kernel from the WOLK project...:). Ugh, don't get me started on WOLK. Anyway, the reason to use non vanilla is that those kernels go through *QA*. > I like 2.4.19 but it isn't a revolution compared to 2.4.18 (meaning - if you > don't need the driver fixes you're ok with either). I agree here. > In general I would recommend making a habit out of installing a vanilla > kernel every time you install a system. This way you can at least report bugs > to LKML. Most people rarely understand how many patches RH applies to their > kernel. I heard a last count of 254 patches most of which wouldn't get by the > first line of kernel maintainers not to mention Linus. Sorry, Mark, I have a rather intimate acquaintaince with RedHat's kernel patches, and most of them find themselves in vanilla and are perfectly fine. Again, the kernel is more than thte sum of its patches, and Redhat's kernels go through QA. For the average desktop user, that's IMPORTANT. Anyway, nice to see you here ;) -- Muli Ben-Yehuda syscalltrack hacker-at-large
msg21949/pgp00000.pgp
Description: PGP signature
