How Linus handles patch requests for linux.

[clrCoda]

It had come to my attention that some folks have interesting and often well intentioned yet possibly misaligned ideas of how an open source project patches code.

I give you Linus' own words about working the kernel repository for possibly, arguably, the most successful open source software project within the realm of the local planet.

Please note that very few people on earth make changes to the linux kernel directly. And yet it is still considered Open Source.

https://www.kernel.org/doc/Documentation/SubmittingPatches

[curt]

Of course I don't speak from a position of always doing everything perfectly or optimally, but I think the above link contains a whole lot of wisdom and truth. It speaks to having realistic expectations, it speaks to the limits/capacity of the human mind when dealing with complexity, it calls for patience when things don't progress as quickly or directly as a person hopes, it puts quite a bit of burden on the patch submitter to carefully craft their patches in a way that helps the code maintainers understand and review them. The linux kernel runs on probably close to half the devices in the world, so there is a responsibility on the part of the code maintainers to proceed carefully and cautiously because any little change can introduce an unexpected bug ... and the consequences of such a bug could be catastrophic if devices start crashing or misbehaving all over the world. There is also implicit recognition by the code maintainers that the existing code is insanely complex and difficult in places, with an insane number of interconnections and subtle nuances that far exceed the capacity of any single person to understand ... so there is great fear and trembling when touching code that hasn't been looked at for a while.

So when I read through the document about how to submit linux patches, I see a whole lot of common sense, a whole lot of wisdom, and a whole lot of recognition of the limits of our human brains ... and a call to be patient with each other's limitations and short comings, because we all certainly have them.

[clrCoda]

"Open Source Code" literally mean that the source code is something any interested party can get his hands on, and typically means that the source is one of the things offered in a distribution of a software package.

From the jargon file : http://www.catb.org/jargon/html/O/open-source.html

It does not necessarily mean open repository. Repository operation is handled much like Linus et. al. spell out, for the reasons that Curt O. has spelled out.

Open Source Software is practical in scope.
Repository operation is logical in scope.

--Ray

[Jabberwocky]

Of course, just from curiosity, I would like to know how you would implement a manually controlled patch system for a gathering of repositories of which about 350 have not even maintainers anymore? Do we need a genie in a bottle taking care of those?
See, here si the difference between a program and a collection of planes. A program has usually all parts maintained and even if something is at a certain day not actively under development, it is probably next week again. At least, that is the reasonable assumption.
In our plane collections, be it FGMEMBERS or FGADDON, we have soem pretty old planes, to valuable to lose them but not maintained in years. We want to keep them in the hope, to resurrect them in the future, but we can't do so right now because we lack more content developers. So they hang there, sometimes years till someone touches them again and does some work on it. In the meantime, FG evolves from version to version. The right thing would be to test fly all of this old planes, check what is wrong, repair them and offer them again to the pilots. The right thing would be to give them maintainers. But where do we get approximate fifty qualified test pilots for the different categories (jets, helis, ...) and another at least thirty more content developers?

[LesterBoffo]

Who get's to authorize who's 'qualified' as a test pilot, or a maintainer.

[wkitty42]

HAHAHAHAH! i was waiting for that question!

[clrCoda]

I believe this thread http://forum.flightgear.org/viewtopic.php?f=4&t=26183&hilit=paper+cut
attempts to try to answer some of those issues.

[Jabberwocky]

Well, basically, the qualification comes with skills, so every volunteer test pilot needs to give some information at least about what kinds of aircraft he is interested in and then we give it a try with some test flights and specific questions? Would that work? I don't know, HerbyW told me, he would organize that, but I can't see him anywhere?

[clrCoda]

I'm not at all sure how this stuff jabberwocky is jabbering ( sorry, couldn't resist hehe ) about relates to this thread.

Post Edit I believe this thread is more on topic for what you speak about here. http://forum.flightgear.org/viewtopic.php?f=4&t=26183&hilit=paper+cut

[IAHM-COL]

Who get's to authorize who's 'qualified' as a test pilot, or a maintainer.

That's the heart of the byz lester

We are all -renegade- Volunteers, and there aint' bosses here

You do as you please.

You can fly any aircraft you want. You can change any aircraft code as you may wish.
You authorize yourself to do exactly as you please. And thus, there is no-one that can fit the shoes of an "aircraft maintainer"
No-one (even original authors) can maintain the posture of "this is MY aircraft" anymore, after the aircraft goes out GPLed.

So basically, test any aircraft you want.
Modify any aircraft you want.

And I, in FGMEMBERs will be looking forward to know those bugs, and debugs, you find or fix. And give them an open, easy wide distribution as well.
Community based efforts.

[Scheiker]

It had come to my attention that some folks have interesting and often well intentioned yet possibly misaligned ideas of how an open source project patches code.

I give you Linus' own words about working the kernel repository for possibly, arguably, the most successful open source software project within the realm of the local planet.

Please note that very few people on earth make changes to the linux kernel directly. And yet it is still considered Open Source.

https://www.kernel.org/doc/Documentation/SubmittingPatches

As a side note, please do not refer to Linux as “Open Source”. The Open Source Initiative defines what “Open Source” is here: http://opensource.org/docs/osd. Linux does not meet the criterion that they defined. Yes the main part of Linux is under GPLv2(a free software license that does meet the criteria of “Open Source”), however there are parts of this kernel that are not under GPLv2 and is in fact many cases non-free proprietary software. How is this possible while using GPL? Well the case is that this non-free software is distributed in the form of firmware that doesn’t get linked with the kernel. Instead it is loaded by the kernel and used in a way that isn’t violating the GPL(otherwise any proprietary software would not be allowed to be used in conjunction with free software). We usually call this type of non-free software “blobs”. The non-free firmware is usually displayed as hexadecimal code cleverly disguised as source code. But hexadecimal code isn’t source code, and thus the source code is not available to you. This doesn’t meet the criteria of “Open Source” by anyones definition, so please don’t call it that(and this perspective is from a free software supporter that is admittedly against many of the values of “Open Source”).

We refer to Linus’s version of Linux as non-free or at least not completely free. The FSFLA(Free Software Foundation Latin America) does produce a kernel that is completely free called “Linux-libre”, see: http://www.fsfla.org/ikiwiki/selibre/linux-libre/. The kernel is produced by running a script that “deblobs” or removes all the non-free parts from the vanilla version of Linux that is released by Linus. It is called “Linux-libre” to distinguish that it is in fact a free version of Linux and not a non-free one. All kernels of free GNU/Linux distros must be deblobed in order to be endorsed by the FSF(Free Software Foundation) and many refer to those free distros as GNU/Linux-libre distros when they use the kernel or script provided by the FSFLA. You can see a list of those distros here: http://www.gnu.org/distros/free-distros.html

In any case, by broadly calling Linus’s version of Linux “Open Source”, you are hurting the primary goals of both free software supporters and “Open Source” supporters.

[clrCoda]

I don't wish to hurt anyone or anything, so thanks for the update

[Scheiker]

I don't wish to hurt anyone or anything, so thanks for the update

No problem , I understand many people don't know about the issues I raised because it is generally not talked about.