Wary is intended to support those on dialup, so I asked myself, what is the most recent kernel that will compile all the dialup drivers?
We have quite a collection of them. I know that half of them don't compile on the 2.6.32 kernel, and they do compile on the 2.6.30 kernel (184.108.40.206 is used in Puppy 4.3.1).
So, that leaves the 2.6.31 kernel to investigate. The most recent is 220.127.116.11 which was released July 2010 -- that is not long ago, so this kernel has been maintained until fairly recently and the changelogs show lots of good fixes have gone into it.
I compiled the 18.104.22.168 kernel last night, then woke up at 2.30am, wasn't sleepy, so tested compiling the dialup drivers -- oh wow, every single one of them compiles!
So, is this "the one" for Wary? Heh heh, I don't know, but it is going to be very interesting to test it. I intend to bring out Wary 0.9.3 in a few days with this kernel.
I was wondering one thing. How difficult would it be to maintain a kernel ourselves? I mean, in the way that the long-term-supported 2.6.27 and 2.6.32 kernels are maintained, by backporting patches. Hmmm...
Note, I have left ALSA at version 1.0.20, as it comes with the kernel. A couple of reasons why I have not compiled the 1.0.23 drivers -- 22.214.171.124 has many ALSA patches so it is really beyond 1.0.20, and version 1.0.20 works well on my test PCs -- including one that "stutters" with ALSA 1.0.23.
Comments:Posted on 26 Oct 2010, 9:28 by ozsouth
Maybe the 126.96.36.199 kernel is suitable for the next Quirky, assuming you go with 188.8.131.52 for Wary.
Posted on 26 Oct 2010, 13:12 by DaveS
Is it possible to put dial-up drivers in a sep. sfs such that those that dont need them dont have to carry the extra size?
Posted on 26 Oct 2010, 15:16 by drongo
Support for dial-up is one of the defining characteristics of Puppy surely? How would those with dial-up download the .sfs?
I haven't used dial-up for at least five years but someone has to support those guys. Up till now it's just been Puppy.
Posted on 26 Oct 2010, 16:23 by Iguleder
It won't be very hard to make our own kernel. We just need three things, a "kernel team", a group of people who have the skills for creating and applying kernel patches, some repository for patches and a coordinator (you!).
We'll have to scan all new kernel changelogs to trace fixes, then we need to find the commits and attempt to apply the fixes on our home-made LTS kernel.
Sounds fun :)
Posted on 26 Oct 2010, 18:16 by linuxcbon
We should not make our kernel. It's too difficult.
Posted on 27 Oct 2010, 9:30 by Raffy
Narrow focus - will this help?
Will a narrow focus in terms of target platforms help in making this possible? Say, focus on low-power platforms like the Intel Atom, VIA C7, and the upcoming AMD Fusion.