| 1 |
On Wed, May 17, 2017 at 7:16 AM, Dale <rdalek1967@×××××.com> wrote: |
| 2 |
> |
| 3 |
> It frustrates me to see Gentoo go through the same issues over and over again with nothing changing. |
| 4 |
|
| 5 |
If you want to fix problems you need to identify their root causes. |
| 6 |
I'm not convinced this is really being done. |
| 7 |
|
| 8 |
Issue: not enough people are contributing to project xyz on Gentoo. |
| 9 |
|
| 10 |
That seems to be the crux of the matter here. Some say it is because |
| 11 |
contributors are scared away by email chains, and that might be |
| 12 |
possibly true. Some say it is because contributors are being turned |
| 13 |
away because they are disliked, and that might even be somewhat true. |
| 14 |
However, I think there are a bunch of other potential causes, and I |
| 15 |
don't think we really can tell how much they all contribute: |
| 16 |
|
| 17 |
First, "not enough people are working on xyz" is a completely |
| 18 |
subjective statement. It is saying that Gentoo does not meet the |
| 19 |
expectations of the person making the statement. There is no |
| 20 |
objective standard that says that a distro must do this much of A and |
| 21 |
that much of B. Every distro has projects that have more and less |
| 22 |
support. A project that one person considers critical might be |
| 23 |
considered superfluous by another. The example in this thread was |
| 24 |
Java and it is a good example of this. I see people with fairly |
| 25 |
divided opinions on Java everywhere. Some deem it an enterprise |
| 26 |
technology that is widespread and indefensible, and their arguments |
| 27 |
are fairly valid. Others call it a memory hog and a pain to |
| 28 |
administer and they usually raise good arguments as well. A lot of |
| 29 |
Gentoo devs don't care about the enterprise, and some do, but probably |
| 30 |
don't use Java there. Also, half the point of Java is |
| 31 |
write-once-run-everywhere which doesn't entirely mesh well with a |
| 32 |
source-based distro. I'm not saying that you can't build Java from |
| 33 |
source - just that you don't necessarily get the same benefits from |
| 34 |
doing so that you might with C. |
| 35 |
|
| 36 |
Then if you want to compare Gentoo with other distros you need to keep |
| 37 |
in mind that we are very much a niche. Many examples can be cited of: |
| 38 |
* Binary distros that are release based. |
| 39 |
* Binary distros that are rolling release. |
| 40 |
* Binary distros that target the enterprise. |
| 41 |
* Binary distros that target the desktop. (ironically the biggest is |
| 42 |
a Gentoo derivative) |
| 43 |
* Binary distros that are focused around Gnome 3. |
| 44 |
* Binary distros that are focused around Gnome 2. |
| 45 |
* Binary distros that are focused around KDE/Plasma. |
| 46 |
* I wouldn't be shocked if there are several focused around KDE 3/4. |
| 47 |
* Binary distros that operate principally from LiveDVDs. |
| 48 |
* Binary distros focused on routers. |
| 49 |
* Binary distros that are very systemd-oriented. |
| 50 |
* Binary distros that use systemd but in more of a legacy/LSB-oriented manner. |
| 51 |
* Binary distros that avoid systemd. |
| 52 |
* Binary distros that exist mainly to run containers (ironically the |
| 53 |
biggest is a Gentoo derivative) |
| 54 |
* Binary distros that exist to run on phones. |
| 55 |
* Binary distros that semi-containerize every package. |
| 56 |
|
| 57 |
If you want to talk about source-based distros there is basically just |
| 58 |
Gentoo and a few derivatives. In the binary world they have such |
| 59 |
manpower available that they can fork themselves in 500 different |
| 60 |
directions. In the source-based world we have so little manpower that |
| 61 |
we struggle to maintain a viable distro under one big tent. |
| 62 |
Maintaining a source-based distro is also fairly manpower-intensive. |
| 63 |
We need build scripts that work for everybody in all kinds of exotic |
| 64 |
configurations. A binary distro just needs to be able to reliably get |
| 65 |
a scripted build to work in a very controlled environment. |
| 66 |
|
| 67 |
I think Gentoo is great, but a lot of people don't feel that they need |
| 68 |
it to meet their needs, and a lot of binary distros are a lot easier |
| 69 |
to maintain. I run OpenWRT on my router, not Gentoo. I run Android |
| 70 |
on my phone, not Gentoo. There are advantages to having |
| 71 |
niche-oriented distros and the binary world has a TON of them. |
| 72 |
|
| 73 |
The binary distros have also tended to improve over time. Back when |
| 74 |
emerge was mainly competing against Debian/Redhat it was a different |
| 75 |
world than the world post-Ubuntu. Arch is also fairly similar to |
| 76 |
Gentoo in principle and thus will tend to split the contributor base. |
| 77 |
|
| 78 |
So, even if we had an environment where every Gentoo contributor was |
| 79 |
completely happy, and every offered PR made it into the tree in a |
| 80 |
quality-controlled manner in a day, we probably would still struggle |
| 81 |
with some of these issues. That is frustrating, but I think it goes |
| 82 |
with the territory. |
| 83 |
|
| 84 |
Don't get me wrong, I'm not trying to belittle the interpersonal |
| 85 |
stuff, and we may be able to improve there. I just wouldn't count on |
| 86 |
it turning into a high-commit-volume paradise without some bigger |
| 87 |
changes. A lot of these changes are going to be difficult to make. |
| 88 |
Our PR workflow is certainly an improvement, but as we see with Java |
| 89 |
it still suffers when nobody with commit access wants to deal with the |
| 90 |
PRs, and there is probably a lot of room for improvement in other |
| 91 |
ways. |
| 92 |
|
| 93 |
-- |
| 94 |
Rich |
Archives