Everyone and their mother creating a patchset

Vidar

I really don't mean to flame here, but it seems like there are 20 different custom kernel patchsets floating around here. I'm not against the idea of bleeding-edge experimental patchsets, but I feel it would be less confusing and probably better for everyone if maybe those that are doing it could work together on a single non-offical "gentoo community" patchset.

Anyone with me on this?
_________________
"Vidar, Odin's mighty son, he will come to slay the wolf
The sword runs into the heart of Hverdrungs son
So he avenges his father" -- Amon Amarth - Burning Creation

Frodg · l33t Joined: 11 Feb 2004 Posts: 761

Where would be the fun in installing an obscure patchset, borking the system and then fixing it?

I would have nothing else to do

_________________
Aerosolo ergo sum - I spray therefore I am

Gentoo - Registered Linux User # 361400

Jake · Veteran Joined: 31 Jul 2003 Posts: 1132

I disagree. Options are good, for example having many distros. Would you like to see Gentoo combined with Fedora, SuSE, and Mandrake? Or worse yet, Lin(dows|spire)? How about combining GNOME and KDE, then throw in the *box wms for good measure?

The "gentoo community" patch is gentoo-sources.

ian! · Posted: Wed Jul 28, 2004 10:58 pm Post subject:

Moved to 'Gentoo Chat' because this is more a chat topic.
_________________
"To have a successful open source project, you need to be at least somewhat successful at getting along with people." -- Daniel Robbins

Roguelazer · Posted: Thu Jul 29, 2004 1:09 am Post subject:

I'm not with you. Why? Because I personally use love-sources, but it's quite buggy usually and I wouldn't reccommend it to somebody running a stable system. Likewise, gentoo-sources are nice (I used to use them), but not bleeding-edge enough for me.

_________________
Registered Linux User #263260

snekiepete · Guru Joined: 07 May 2003 Posts: 306 Location: WISCONSIN

With there being so many different choices for so many different things, it would be hard to restrict kernel sources to limited versions, because there are different cpu schedulers, file systems, etc....this patch that does this and this patch that does that, the choice is a blessing, and for those who are not skilled in kernel patching, I am sure they enjoy the choices. I myself patch my own stuff.

Arker · Apprentice Joined: 10 Sep 2002 Posts: 205

I think it's fantastic that this sort of thing happens in the Linux community. To me, that is what it's all about!

~djc
_________________
*LIK*

My other computer is your Windows box.

Pink · Veteran Joined: 24 Jul 2003 Posts: 1062

I think you have a naiive outlook on the patchsets.

If you actually read the patchset notes with each one they are very different.

Take just two, love-sources and nitro-sources. I'm sure you know but they include some of the latest patches but they differ significantly in that one prefers the np scheduler the other the staircase scheduler. Now how exactly would you combine those? Especially now that love-sources has slimmed down and gives you the option of adding some extra patches if you want, whereas nitro is preferable, IMO, because it combines it all in one good patch.

Other patches are even more bleeding edge and others concentrate on different types of sched to choose from. Others are based on the mm-patchset others are based on ck.

All of which are significantly different in their base structure and what they actually offer the individual.

What exactly is your idea for one patchset to combine them all - considering you cannot have more than one sceheduler in a kernel, and some people would want bleeding edge patches on bleeding edge bases whilst others simply want new patches (for officially unsupported hardware) on a good up-to-date base? What would you do?

Vidar · Posted: Thu Jul 29, 2004 7:32 am Post subject:

Alright, you have a good point....

I was thinking that the best way to go would be to specify (at config time, use flags, whatever) things like schedulers or whatever else is a one-or-the-other thing. But is that even possibile?

Still, however, I will say that the amount of custom patchsets floating around is confusing. Maybe the proper way to go would be to have some sort of an faq with latest versions / changelogs, information about the aim of each patchset so then the adventurous user can select one of them without having to dig through all the individual patches to see what the hell is going on.
_________________
"Vidar, Odin's mighty son, he will come to slay the wolf
The sword runs into the heart of Hverdrungs son
So he avenges his father" -- Amon Amarth - Burning Creation

seppe · Posted: Thu Jul 29, 2004 2:58 pm Post subject:

Why are more patchsets not good? GNU/Linux is all about choice, and more patchsets gives you more choice

I used gentoo-dev-sources first, then I switched to love-sources to enjoy the 'bleeding edge feeling', and now I created my own patchset (nitro-sources) because I like the scheduler of CK (Staircase) more than Nick Piggin's scheduler (which is used in love-sources) + I added other pathces which seems to be popular (reiser4, bootsplash, win4lin, lirc, supermount, ...).
Thanks to this, people have more choice now .. and that's a good thing!

Would you like it that Gnome, KDE, XFCE and FluxBox became integrated in 1 big slow DE? I don't think so

And if you don't like the available patchsets, then create your own!

_________________
nitro-sources, because between stable and experimental there exists only speed

Latest release I made: 2.6.13.2-nitro1

Titeuf · Posted: Thu Jul 29, 2004 4:00 pm Post subject:

There are never too many patchsets for the kernel. There must be one for the needs of everybody.
But one thing I notice with Gentoo is that more people are testing new/unstable/bleeding-edge things.

codergeek42 · Posted: Thu Jul 29, 2004 9:53 pm Post subject: completely OT...

Wait, wait....wait ....

My mother is creating a kernel patchet?

_________________
~~ Peter: Programmer, Mathematician, STEM & Free Software Advocate, Enlightened Agent, Transhumanist, Fedora contributor
Who am I? :: EFF & FSF

PrakashP · Posted: Thu Jul 29, 2004 9:58 pm Post subject:

Oh guess what, despite all the patch-sets floating around, I am keeping my own personal kernel patch-set...

Ateo · Posted: Mon Aug 02, 2004 9:21 pm Post subject:

I'll ask my mom what patchset she's created......

fallow · Posted: Sun Aug 08, 2004 10:18 pm Post subject:

Is many of "patchsets" at now ...more patchsets = better choice . The choice is the one of "these" important things in ascpects of using linux.

I think is never to much

I prefer single prior table shedulers - staircase and spa hybrid and also runtime elevation of cpu scheduler so I`m using my own with Peter William`s Hydra scheduler and is many of examples and ways to choose "patchset" for yourself.If somebody feel that no one is good , then creating new

and this is very good IMO

xx-sources is good also

greetings

_________________
"Time is a companion that goes with us on a journey. It reminds us to cherish each moment, because it will never come again. What we leave behind is not as important as how we have lived" J-L. Picard

Thalion · Posted: Sat Aug 14, 2004 12:39 pm Post subject:

All those different patches... it sounds too much like those binary packages, which, as we know, have been superceded by Gentoo USE-flags =) Now, we only need a single kernel source, and flags to enable/disable various patches!

On a serious side, could such a thing really be done? Obviously, many patches are mutually exclusive, so such an ebuild would have to have lots of rules to filter out invalid combinations. But if you just think about the possibilities... and so many more ways to break your system ;)

Sasuke2k · n00b Joined: 14 Aug 2004 Posts: 1

Flaming rocks!

pilla · Posted: Sat Aug 14, 2004 3:44 pm Post subject:

mother-sources, just using the 2.2 branch of kernels and without a very good performance -- but very safe!

the kernel would avoid any attempt of playing games or going online for too many hours.
_________________
"I'm just very selective about the reality I choose to accept." -- Calvin

Archangel1 · Veteran Joined: 21 Apr 2004 Posts: 1212 Location: Work

playfool · Posted: Sun Aug 15, 2004 10:11 am Post subject:

Every vendor has it's own patchset, every major developer has his/her own patchset (most are not released to the public though) - a natural extention of this is that every advanced user wants to extend his/her vpenis by making his/her own patchset - the only problem is that most of these new patchsets contain no new developments, nor stability testing - thus they are pretty much useless in the large context.

it bothers me, I wish vanilla was good enough for everyone to use it, it would be easier to manage and the kernel would get much wider testing and development.

apeitheo · Apprentice Joined: 09 Jan 2004 Posts: 222

butters · Posted: Mon Aug 16, 2004 12:58 am Post subject:

Instead of merely registering my utmost support for kernel hackers and patchers alike (Alan Cox, Andrew Morton, and countless others started off with their patchsets of fixes and features, after all), I offer a suggestion for "validating" these kernel patchsets to ensure quality.

It seems like the only valid complaint with the proliferation of patchsets is the tendency towards reduced quality, stability, and consistency that goes along with shoving rejected patch hunks into the kernel a few times per week. It seems like the kernel has evolved to the point where verification is necessary, and here's how:

Every patch must come with a simple test suite that provides a way of testing the functionality of that patch. This should also apply the the functional blocks of the mainline kernel. After running make on the kernel sources, the user should be able to issue a "make check" command that boots the kernel within a supplied (and sandboxed) virtual machine and runs this series of tests on the kernel image. In particular, all filesystems should test their ability to mount, read, write, and unmount their respective volumes; the schedulers should test their ability to spawn and schedule processes; and a check for gross memory leaks should be done.

The implications of this are manifold. Server admins will be able to test their new kernels before taking their machine down for reboot. Kernel hackers can instantly check if their changes have had effects on other kernel subsystems. Corporate CEO's can quit their blood-pressure medication, knowing that their next security update won't take down their network. And enthusiasts trying their hand at patchsets can make sure that win4lin works correctly with CK's low latency patches.

This is obviously a big architectural undertaking, mandating the involvement of the entire core kernel team, but a self-checking kernel would be grounds for a 2.7 branch, and probable cause for a Linux 3.0. Since this has been moved to Chat, what do you guys think?
_________________
If tugboats were bigger, they'd be the ones getting tugged.

Billy_Witchdoctor · n00b Joined: 11 Aug 2004 Posts: 10

Pink · Veteran Joined: 24 Jul 2003 Posts: 1062

FormerSlacker · Posted: Tue Aug 17, 2004 9:24 pm Post subject:

Not to knock anyone, but personally, unless they are kernel programmers I wouldn't trust them to patch the kernel and fully understand the consequences of applying patch x y and z.

Then again I'm a prude and just stick with the vanilla sources.