| View previous topic :: View next topic |
| Author |
Message |
mimosinnet
l33t
Joined: 10 Aug 2006
Posts: 713
Location: Barcelona, Spain
|
 Posted: Wed Jul 02, 2014 7:49 am Post subject: |
 |
|
I have been using gentoo for some years now. I have been leading towards a light system using low resources (fvwm). Despite of this, each upgrade involved long emerge times (like dev-qt/qtwebkit). I tried ArchLinux and, despite it was a good experience in terms of fast installation, the total configuration and control of the system was somehow lost. Also, I have several gentoo boxes and I like to have the same configuration in all boxes despite being different systems. I have now a binhost server that produces the binaries for my different boxes, having the best of both worlds. Somehow, I feel like having "my own" distribution.
Cheers!
_________________
Please add [solved] to the initial post's subject line if you feel your problem is resolved.
Take care of the community answering unanswered posts. |
|
| Back to top |
|
 |
developer1
n00b
Joined: 29 Mar 2014
Posts: 49
Location: PL
|
| Posted: Fri Jul 11, 2014 1:16 pm Post subject: |
|
|
| Perfect Gentleman wrote: | | i've been using Arch Linux for couple of years, and don't have any problems with it. |
same here......  Its far better (and far more stable) than Gentoo...... |
|
| Back to top |
|
|
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
| Posted: Fri Jul 11, 2014 10:05 pm Post subject: |
|
|
| Budoka wrote: | | krinn wrote: | It also depends on what you call "breakage".
If having portage complain for slot conflict is a breakage, yep, i understand your frustration then.
But for me breakage is when portage is doing the work, but the result of the work is a dead program or system. And after all those years i'm sure the breakage number is still below 10! (by memory i remember e2fsprogs update breaking wget and mpfr update breaking gcc only). |
Yes that is indeed what I mean. I have only had portage bork my system once and I was able to get it fixed quite quickly. |
Well, you'll get the other type of breakage on other distros.
This is an important difference, and something people who don't run Gentoo don't realise; when Gentoo users talk about breakage, it usually means they can't get past some conflict, and they're talking about "breakage" in the portage tree.
Yes we get just as much "breakage" as other distros; but in the meantime everything carries on working. Breakage on bindists only means the type of breakage where your software simply won't work (or more often fails in very mysterious ways, depending on what libs are currently installed.)
And sure, we come across bugs, like everywhere else. But I for one have had a much more stable time on Gentoo than anywhere else.
I wouldn't ever give that up, certainly not to go back to the weird dependency hell of various libs frozen in someone else's configuration at some point in time, reinstalling every 6 months to get a new set of weirdness.
From what I read, arch has effectively no QA; the Gentoo stable/unstable tree split works very well imo. |
|
| Back to top |
|
|
gerard27
Advocate
Joined: 04 Jan 2004
Posts: 2377
Location: Netherlands
|
| Posted: Fri Jul 11, 2014 11:05 pm Post subject: |
|
|
steveL ++
Gerard.
_________________
To install Gentoo I use sysrescuecd.Based on Gentoo,has firefox to browse Gentoo docs and mc to browse (and edit) files.
The same disk can be used for 32 and 64 bit installs.
You can follow the Handbook verbatim.
http://www.sysresccd.org/Download |
|
| Back to top |
|
|
hasufell
Retired Dev
Joined: 29 Oct 2011
Posts: 429
|
| Posted: Wed Jul 16, 2014 12:12 am Post subject: |
|
|
| developer1 wrote: | | Perfect Gentleman wrote: | | i've been using Arch Linux for couple of years, and don't have any problems with it. |
same here...... Its far better (and far more stable) than Gentoo...... |
We can go this way if you want.
I come from arch linux, have probably used it for almost 2 years as well and were also actively doing some stuff on AUR (since arch is basically useless without AUR... and that's not a positive thing).
I have never felt that it is stable. The PM is very quick, but that's because it almost does nothing (like conflict resolution). So... for big updates, you have way more trouble than I ever had with gentoo updates and it's way more easy to break stuff. You always had to remember: _always_ read the news before updating. There might be a very important hint to save you from a broken system.
The times where this was necessary in gentoo can probably be expressed with a boolean.
So really... either this has drastically improved since last I checked or you have just a low-maintenance system. |
|
| Back to top |
|
|
davidm
Guru
Joined: 26 Apr 2009
Posts: 557
Location: US
|
| Posted: Sat Jul 19, 2014 5:41 pm Post subject: |
|
|
| hasufell wrote: | | developer1 wrote: | | Perfect Gentleman wrote: | | i've been using Arch Linux for couple of years, and don't have any problems with it. |
same here...... Its far better (and far more stable) than Gentoo...... |
We can go this way if you want.
I come from arch linux, have probably used it for almost 2 years as well and were also actively doing some stuff on AUR (since arch is basically useless without AUR... and that's not a positive thing).
I have never felt that it is stable. The PM is very quick, but that's because it almost does nothing (like conflict resolution). So... for big updates, you have way more trouble than I ever had with gentoo updates and it's way more easy to break stuff. You always had to remember: _always_ read the news before updating. There might be a very important hint to save you from a broken system.
The times where this was necessary in gentoo can probably be expressed with a boolean.
So really... either this has drastically improved since last I checked or you have just a low-maintenance system. |
I've bounced back and forth between both Gentoo and Arch for a while now. About four or five years. I've been mainly using Arch for the past couple years. To be honest I haven't had any real problems recently with Arch. But what you said about problems with the dependencies is sometimes true. Things will install but sometimes it doesn't work. It's very, very rare though. At least for me.
Gentoo OTOH I often had problems compiling things and I got sick of the constant breakage (it may have been a bad time or tha tI did not learn all the tricks yet). At the time I only had a P4 system so constant compiling was a major pain because my system was already overtaxed as it was and it meant I could do little while compiling as far as work goes.
Yet at the moment I'm now considering installing Gentoo again because Arch Linux almost has me too bored. Nothing goes wrong and it's very easy. Ho-hum. 'Pacman -Syu' and it just works. With Gentoo I got to have a little fun and learn a bit in the process. Oh well. I might install Gentoo on my old P4 again to satisfy that itch. |
|
| Back to top |
|
|
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
| Posted: Mon Jul 21, 2014 4:56 pm Post subject: |
|
|
| davidm wrote: | | hasufell wrote: | I come from arch linux, have probably used it for almost 2 years as well and were also actively doing some stuff on AUR (since arch is basically useless without AUR... and that's not a positive thing).
I have never felt that it is stable. The PM is very quick, but that's because it almost does nothing (like conflict resolution). So... for big updates, you have way more trouble than I ever had with gentoo updates and it's way more easy to break stuff. You always had to remember: _always_ read the news before updating. There might be a very important hint to save you from a broken system.
The times where this was necessary in gentoo can probably be expressed with a boolean.
So really... either this has drastically improved since last I checked or you have just a low-maintenance system. |
I've bounced back and forth between both Gentoo and Arch for a while now. About four or five years. I've been mainly using Arch for the past couple years. To be honest I haven't had any real problems recently with Arch. But what you said about problems with the dependencies is sometimes true. Things will install but sometimes it doesn't work. It's very, very rare though. At least for me. |
Well what I referred to above as "from what I've read" was this thread. (See siljrath and PhaytalError's posts.) Sounds like you've been lucky though.
| Quote: | | Gentoo OTOH I often had problems compiling things and I got sick of the constant breakage (it may have been a bad time or tha tI did not learn all the tricks yet). At the time I only had a P4 system so constant compiling was a major pain because my system was already overtaxed as it was and it meant I could do little while compiling as far as work goes. |
Yeah though as pointed out "breakage" in Gentoo is a world away from the true breakage you get in bindists. At least ime, running a stable system, and only keywording packages I'm happy to deal with breaking. I could never go back to true breakage, usually stuff where you have no clue what's broken in the massive stack of layer-upon-layer. And ofc it never stays still, as you want to stay current with bugfixes. In contrast I can take up to a year out of updating Gentoo if I need to, and I'm happy with how everything is running.
I only do this when I have current "breakage" (ie slot-conflict, or toolchain package hard-masked) revealing a situation update should be handling, and the only times that have been that long were expat (original /etc/warning ABI upgrade, which led to work in chroots and multiple-binhosts) and currently as I want to finish --toolchain, and perl came in with a warning, which needs to be handled there too. It's taken so long as this has also led to other complications like filtering conflicts (4 types, last one hardmasking still to add) since a package we have a warning for may be blocked, and we want to continue where we'd normally bail out (only one, dep conflicts iirc, doesn't usually lead to an error from portage.) And ofc the actual thing I want to work on has been mostly done, but is delayed by handling toolchain warnings, which may also be blocked.
Sorry for rambling: my point was that I used to run a 32-bit install for many years, and because I was working on update, I tended to do lots of compiles. (It's one big long script apart from the lib, as I somewhat perversely wanted to see how far I could stretch bash. I never got it to fall over, before I finally went to amd64.) Being able to use pkgcore instead of portage made the biggest difference, since once you've set it off, it tends to run in the background (and pmerge is very quick; it's like the difference between gentoolkit and portage-utils, or esearch and eix.) I can't wait for pkgcore to be updated, now that someone's working on it again.
Session cgroups (kernel option) help with that, though I didn't use them back then.
| Quote: | | Yet at the moment I'm now considering installing Gentoo again because Arch Linux almost has me too bored. Nothing goes wrong and it's very easy. Ho-hum. 'Pacman -Syu' and it just works. With Gentoo I got to have a little fun and learn a bit in the process. :) Oh well. I might install Gentoo on my old P4 again to satisfy that itch. :) |
I'm on a fairly old core2 duo workstation, which I only upgraded to 4G RAM from 2G about a year ago. Once you've got things how you like them, you can run a very quick machine. In my case, it was getting rid of semantic-desktop in KDE-4 which made the biggest difference. Though I had to give up KMail which hurt, and it took a while to find out what supported maildir properly, switching to mutt taught me postfix which I've always wanted to know; and now I love it, as it's so quick and lightweight. Plus it meant I could get rid of nubkit too.
Though I've always tinkered with my install, and I love that aspect of Gentoo, the real advantage for me is stability. I spend my time using the machine, instead of wrestling with the distro.
Once you're comfortable with Gentoo, it's rock-solid by comparison to every Linux distro I've used, without locking me into someone else's choices, and bugs. |
|
| Back to top |
|
|
The Doctor
Moderator
Joined: 27 Jul 2010
Posts: 2678
|
| Posted: Mon Jul 21, 2014 5:58 pm Post subject: |
|
|
The only time I have ever had stability problems on Gentoo was when I was mixing stable and unstable branches. Some times overlay software can be unstable, but that is to be expected.
On the other hand, Arch was so unstable that I stopped using it shortly after they switched to systemd. I had been using it on a laptop to avoid compiling times and wear, but the resulting install proved to be too unstable to use. One update literally erased /etc/shadow (as best I could determine) so I replaced it with Gentoo. Quite simply, if I'm working from a laptop I can't have it failing to work.
I think the real question is what you want to use Linux for. If you like to tinker and want/need stability you can't beat Gentoo. If you just need stability then maybe Debian or Slackware would fit best. For ease of use something like Mint is appropriate. If you absolutely need to be on the bleeding edge then Arch may be your ticket, but at the cost of stability and usability.
_________________
First things first, but not necessarily in that order.
Apologies if I take a while to respond. I'm currently working on the dematerialization circuit for my blue box. |
|
| Back to top |
|
|
Cyker
Veteran
Joined: 15 Jun 2006
Posts: 1746
|
| Posted: Mon Jul 21, 2014 7:19 pm Post subject: |
|
|
Gentoo's strength and weakness is its rolling update nature; Eventually you'll get it to a state that you want to keep but unless you stop updating (--sync'ing) then you will not be able to stay there for too long.
Also, where other distros would require you to install a totally new version every so often, Gentoo will never do that, and allow you continue with your existing system, even through hardware changes (My old 32-bit system started as a humble P200MMX and is now a an Opteron 180!). However, the downside of this is occasionally you get hugely disruptive updates which will require a great deal of care to sort out. Sometimes this is because of fundamental changes in the core of (e.g. gcc or libc; always scary!) but I've had random useflag changes which have caused huge useflag cascades which suddenly want to pull in hundreds of megs of packages suddenly.
Gentoo can do almost anything you could possibly want or need it to, but the one thing you need more than anything else to use it effectively is LOTS OF TIME!!! |
|
| Back to top |
|
|
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
| Posted: Tue Jul 22, 2014 4:40 pm Post subject: |
|
|
| I don't get the "rolling update" issue, myself. I used to update far more often when I used bindists. |
|
| Back to top |
|
|
deefster
Tux's lil' helper
Joined: 19 Apr 2004
Posts: 77
|
| Posted: Wed Jul 23, 2014 9:04 am Post subject: |
|
|
| I take LVM snapshots of my root logical volume daily. This allows for rollback should any major breakage occur. Breakage doesn't just have to be from package compilation or dependancy issues. It can also be config file changes, maybe a big change in xorg or switching to a new init system, etc. I sync portage, fetch packages, and send myself the pretend upgrade output daily, and based on the scope of that effort, I may create an upgrade snapshot of my root partition, chroot into that to install the updates. Once everything is compiled, I can even boot into it with a grub boot option, and ensure everything works ok before merging it back into my stable environment. |
|
| Back to top |
|
|
hasufell
Retired Dev
Joined: 29 Oct 2011
Posts: 429
|
| Posted: Wed Jul 23, 2014 1:01 pm Post subject: |
|
|
| Cyker wrote: | | Gentoo's strength and weakness is its rolling update nature |
The weakness of gentoo is its organizational structure, which is the main cause for the broken workflow and general lack of orientation.
If gentoo fails, then because of that. |
|
| Back to top |
|
|
khayyam
Watchman
Joined: 07 Jun 2012
Posts: 6227
Location: Room 101
|
| Posted: Wed Jul 23, 2014 2:14 pm Post subject: |
|
|
| hasufell wrote: | | The weakness of gentoo is its organizational structure, which is the main cause for the broken workflow and general lack of orientation. If gentoo fails, then because of that. |
hasufell ... I'm not sure what specific aspects of the organisational structure you mean but I would generally agree. It seems that there is a lot in place (the council, herds, Q&A, comrel/devrel, etc) but that these are mostly reactive (in that the tend to take action, or come into play, after the fact). The basic premise is that the philosophy and CoC (with some governance structure in place ... the council, comrel/devrel, etc ... to adjudicate, make decisions, etc) is sufficient. Such an adhoc arrangement can work (in cases where there is consensus and the size of the particular group is small), but it leaves a lot to be desired when it comes to scale ... and so direction, interrelations, working methods, etc, tend to suffer. I might hazard that the reason behind the call for a council "president" is a result of this, though I don't think it'd offer much in the way of a solution (in fact its more likely to exacerbate things).
What I think is needed is a more formalised system of rules and a method by where these rules, and the actions resulting from these rules, can be evaluated (something of a senate/council consisting of representative elected by their constituents ... and re-callable by same). This senate/council needs to be represented by all parties and should be proportional (it might also be a good idea to include some representatives from the user community), it should not need to convene for every decision only substantive questions of governance (in other words it shouldn't get bogged down in minutiae). I recognise that such a system would require a lot in terms of resources (and I'm no doubt unrealistic in expecting it might be possible to achieve) but in organisational terms I think something of this nature is needed.
best ... khay |
|
| Back to top |
|
|
hasufell
Retired Dev
Joined: 29 Oct 2011
Posts: 429
|
| Posted: Wed Jul 23, 2014 4:05 pm Post subject: |
|
|
| khayyam wrote: | | hasufell wrote: | | The weakness of gentoo is its organizational structure, which is the main cause for the broken workflow and general lack of orientation. If gentoo fails, then because of that. |
hasufell ... I'm not sure what specific aspects of the organisational structure you mean but I would generally agree. It seems that there is a lot in place (the council, herds, Q&A, comrel/devrel, etc) but that these are mostly reactive (in that the tend to take action, or come into play, after the fact). The basic premise is that the philosophy and CoC (with some governance structure in place ... the council, comrel/devrel, etc ... to adjudicate, make decisions, etc) is sufficient. Such an adhoc arrangement can work (in cases where there is consensus and the size of the particular group is small), but it leaves a lot to be desired when it comes to scale ... and so direction, interrelations, working methods, etc, tend to suffer. I might hazard that the reason behind the call for a council "president" is a result of this, though I don't think it'd offer much in the way of a solution (in fact its more likely to exacerbate things).
What I think is needed is a more formalised system of rules and a method by where these rules, and the actions resulting from these rules, can be evaluated (something of a senate/council consisting of representative elected by their constituents ... and re-callable by same). This senate/council needs to be represented by all parties and should be proportional (it might also be a good idea to include some representatives from the user community), it should not need to convene for every decision only substantive questions of governance (in other words it shouldn't get bogged down in minutiae). I recognise that such a system would require a lot in terms of resources (and I'm no doubt unrealistic in expecting it might be possible to achieve) but in organisational terms I think something of this nature is needed.
best ... khay |
There are just too many developers with too diverging ideas and concepts. Some old-timer developers like to propagate that this is good for gentoo and that conflicting projects are ok. Experience tells me it is not, despite the council being in place.
Gentoo relies heavily on centralized packaging and decentralized mechanisms don't work well because of the way "profiles/" works under gentoo. In addition our workflow with bugzilla and CVS is so terrible that most potential contributors just refrain from becoming gentoo devs (and I have asked a lot) and rather use a github based overlay.
Also interesting read from an exherbo developer and former gentoo dev.
| Quote: |
The project website says that "current source based distributions were a good starting point. However, to provide what we need and what we want, a new generation of tools and development is required." Which kinds of tools are missing in other source based distributions?
Other source based distributions follows a fairly closed development model that relies on a particular group of developers doing most, if not all the work and a somewhat complex organisation model that's supposed to help solve internal problems. The most common solution when technical problems (such as packages not getting timely updates) occurs is to add more developers to the organisation. Unfortunately this also tends to amplify any organisational problems.
So I'm going in a totally different direction with Exherbo and deliberately trying to keep the amount of developers minimal and instead make sure that we have the tools and infrastructure needed for a much more distributed model. This way we can keep the core team small and make sure we agree on the direction of Exherbo and still benefit from a big number of contributors. |
That is basically what the linux kernel does. It has very few core developers who touch the hard stuff and a lot of ideas are just declined because they don't go the way the core developers want. Still, they have a shitload of random contributors.
Gentoo today is a bunch of loosely grouped people who all have a different understanding of what gentoo is, randomly working together or against each other, all (supposedly) on the same level. Communication is either non-existent, broken or only works in very small groups. The council tries to keep the status quo and make the thing not fall apart. It does not give new ideas, directions or whatever. Most of the council members (with very few exceptions) don't even get involved in open discussions and don't try to mediate. Some of them seem idle since years (just picked a random one), almost never show up on any mailing list and bluntly repost their old manifesto. The power in gentoo comes from the old-timers.
We lack professionalism on a lot of fronts and put people into positions they are just not fit for. Yes, I don't care if that is offensive. It can cause damage.
IMO, there is no way to fix a system that is in such a state. All we can do is not make it worse. But we are on that track already.
If you want you could say that gentoo is broken technically (I skipped that large part about portage, eclasses and other things), organizationally and socially. Still, it works for a lot of people in a very weird way and I don't see sensible alternatives that don't suck 200 times more, yet. |
|
| Back to top |
|
|
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
| Posted: Thu Jul 24, 2014 3:33 pm Post subject: |
|
|
| hasufell wrote: | There are just too many developers with too diverging ideas and concepts.
Gentoo today is a bunch of loosely grouped people.. The power in gentoo comes from the old-timers. |
This sounds exactly the same as:
| Quote: | | That is basically what the linux kernel does. It has very few core developers who touch the hard stuff and a lot of ideas are just declined because they don't go the way the core developers want. Still, they have a shitload of random contributors. |
though a little more distributed, since it's not dependent on a BDFL any more.
Linux has a similar core of people, same as the "old-timers" in Gentoo.
| Quote: | | We lack professionalism on a lot of fronts and put people into positions they are just not fit for. Yes, I don't care if that is offensive. It can cause damage. |
Well that's certainly true. I find it amazing that any random kid can sign up to be in the QA team for example. Though ime, the best and only QA are bug-wranglers; I don't consider anyone else even to be in the QA team.
This also relates to "that large part about portage, eclasses and other things"; big changes like multilib and python targets should be worked on in overlay, proven to work and tested, long before they hit the tree.
I think the problem is lack of experience, coupled with a determination not to be told what to do, which means it takes two years to get across a point that could have been grasped in a week at the beginning, if it'd been thought about instead of rejected as alien. The most telling experience is realising how little you know, which is what makes collaboration (and normal development) so much fun, when done right. You're always learning new stuff from people who know the domain better.
| Quote: | | Still, it works for a lot of people in a very weird way and I don't see sensible alternatives that don't suck 200 times more, yet. |
Indeed: I'd argue you could say the same about FLOSS in general, and most of your points are really about that.
| Quote: | | Some old-timer developers like to propagate that this is good for gentoo and that conflicting projects are ok. Experience tells me it is not, despite the council being in place. |
Conflicting packages and indeed in-house projects to implement part of the toolchain or base-system are ofc fine. But for that to work there needs to be an understanding, and acceptance, of the underlying principles that got us here. Diversity in an ecosystem is always stronger, and always wins out in the end. Without it you cannot have competing projects, which makes testing and improvement of components a lot less capable (since you can't swap them in and out to compare, contrast and eliminate as a problem.)
| Quote: | | Gentoo relies heavily on centralized packaging and decentralized mechanisms don't work well because of the way "profiles/" works under gentoo. In addition our workflow with bugzilla and CVS is so terrible that most potential contributors just refrain from becoming gentoo devs (and I have asked a lot) and rather use a github based overlay. |
Well flattening the profiles to make the basedirs effectively mixins definitely needs to happen, and I agree that using a portage gittree would make the backend work a lot easier. But from a user/admin pov, it makes no odds; it's still better to rsync the deployment end, and in product terms that's what matters.
People collaborating in overlays is exactly the purpose they were intended for. So in effect we have many contributors, all communicating excellently via vcs, and then pushing to a central core of 100-150 developers. Sounds pretty good to me. |
|
| Back to top |
|
|
hasufell
Retired Dev
Joined: 29 Oct 2011
Posts: 429
|
| Posted: Thu Jul 24, 2014 6:57 pm Post subject: |
|
|
| steveL wrote: | | People collaborating in overlays is exactly the purpose they were intended for. So in effect we have many contributors, all communicating excellently via vcs, and then pushing to a central core of 100-150 developers. Sounds pretty good to me. |
That's not reality.
Most of what happens in overlays never ends up in the tree, because no one knows and no one communicates. Even projects like science that do have a relatively strong userbase contributing to their official overlay... rarely import that stuff into the tree.
Sunrise is dying too. Our collaboration channels have basically failed. Our system is not designed for it either, which is why we keep adding developers and make things worse.
| steveL wrote: | | Indeed: I'd argue you could say the same about FLOSS in general, and most of your points are really about that. |
Not all projects mess up as bad as gentoo. No, definitely not. |
|
| Back to top |
|
|
ulenrich
Veteran
Joined: 10 Oct 2010
Posts: 1480
|
| Posted: Thu Jul 24, 2014 8:10 pm Post subject: |
|
|
| hasufell wrote: | | Gentoo relies heavily on centralized packaging and decentralized mechanisms don't work well because of the way "profiles/" works under gentoo. In addition our workflow with bugzilla and CVS is so terrible that most potential contributors just refrain from becoming gentoo devs (and I have asked a lot) and rather use a github based overlay. |
Maybe my ideas about what Gentoo should mean are similar. I think:
- Gentoo should be a pure meta-distribution!
- Gentoo should not guide end users (do not provide full featured stable and testing releases!)
- Gentoo instead should provide advanced users the means to provide many releases.
Portage as a tool has a lot of means! Gentoo as it is now presenting itself, hides all of these.
- Abandon recursive profiles!
- Introduce the use of /cate-gory/release-NAME.ARCH lists instead of keywords! (*)
- Use a much more simpler way of just "positives" instead of all these masks in the profiles!
(*) these release-NAME.ARCH lists should reside in an extra tree. Thus end-user releases are little external repositories of a handful of /cate-gory/release-NAME.ARCH lists and a profile with USEs. As emerge will only find the listed files, this will hurry up emerge as a side effect.
Last edited by ulenrich on Thu Jul 24, 2014 8:21 pm; edited 1 time in total |
|
| Back to top |
|
|
hasufell
Retired Dev
Joined: 29 Oct 2011
Posts: 429
|
| Posted: Thu Jul 24, 2014 8:17 pm Post subject: |
|
|
| ulenrich wrote: | | hasufell wrote: | | Gentoo relies heavily on centralized packaging and decentralized mechanisms don't work well because of the way "profiles/" works under gentoo. In addition our workflow with bugzilla and CVS is so terrible that most potential contributors just refrain from becoming gentoo devs (and I have asked a lot) and rather use a github based overlay. |
Maybe my ideas about what Gentoo should mean are similar. I think:
- Gentoo should be a pure meta-distribution!
- Gentoo should not guide end users (do not provide full featured stable and testing releases!)
- Gentoo instead should provide advanced users the means to provide many releases.
Portage as a tool has a lot of means! Gentoo as it is now presenting itself, hides all of these.
- Abandon recursive profiles!
- Introduce the use of /cate-gory/release-NAME.ARCH lists instead of keywords! (*)
- Use a much more simpler way of just "positives" instead of all these masks in the profiles!
(*) these release-NAME.ARCH lists should reside in an extra tree. Thus end-user releases are little external repositories of a handful of /cate-gory/ lists and a profile. |
All interesting ideas, yet none of it will happen.
The last guy who was actively working on GLEPs with non-trivial changes to ebuild language and other stuff was ferringb. He ragequitted. |
|
| Back to top |
|
|
ulenrich
Veteran
Joined: 10 Oct 2010
Posts: 1480
|
| Posted: Thu Jul 24, 2014 8:37 pm Post subject: |
|
|
| hasufell wrote: | | ulenrich wrote: | | hasufell wrote: | | Gentoo relies heavily on centralized packaging and decentralized mechanisms don't work well because of the way "profiles/" works under gentoo. In addition our workflow with bugzilla and CVS is so terrible that most potential contributors just refrain from becoming gentoo devs (and I have asked a lot) and rather use a github based overlay. |
Maybe my ideas about what Gentoo should mean are similar. I think:
- Gentoo should be a pure meta-distribution!
- Gentoo should not guide end users (do not provide full featured stable and testing releases!)
- Gentoo instead should provide advanced users the means to provide many releases.
Portage as a tool has a lot of means! Gentoo as it is now presenting itself, hides all of these.
- Abandon recursive profiles!
- Introduce the use of /cate-gory/release-NAME.ARCH lists instead of keywords! (*)
- Use a much more simpler way of just "positives" instead of all these masks in the profiles!
(*) these release-NAME.ARCH lists should reside in an extra tree. Thus end-user releases are little external repositories of a handful of /cate-gory/ lists and a profile. |
All interesting ideas, yet none of it will happen.
The last guy who was actively working on GLEPs with non-trivial changes to ebuild language and other stuff was ferringb. He ragequitted. |
You can do it without any GLEPs :
a) Just fork emerge to let it just acquire the ebuilds from release lists, ignoring all of the keywords.
b) Anyone who wants to learn about portage profiles, use this script to serialize a profile:
| Code: | #!/bin/bash
P="/etc/portage/make.profile"
T="target"
#@
#@ purpose:
#@ generate a profile in /etc/portage/make.profile
#@ like as if it is a /etc/portage/make.profile.d
#@
#@ pre requisite:
#@ /etc/portage/make.profile must NOT be a link
#@ /etc/portage/make.profile must be a directory containing
#@ /etc/portage/make.profile/target must link to a portage/profile
function removeOldStackDirs () {
cd "$P" || exit
for i in *.$T.enable ; do
[ -d "$i" ] || continue
cd $P/$i || exit
for j in * ; do
[ -L $j ] && rm "$j"
done
cd "$P"
rmdir "$i"
done
}
function makeStackDirs () {
# recursive function!
pushd "$1" 1>/dev/null || exit
if [ -f "./parent" ] ; then
#instead tac "$PWD/parent" ?
while read c ; do
d=${c/*:}
[ "$c" != "$d" ] && c="${PBase}/$d"
makeStackDirs "$c"
done < ./parent
fi
N=$(( $N + 10 ))
y="$(basename $PWD)"
x="$(basename ${PWD/?$y})"
mkdir -p "$P/$N-$x-$y.$T.enable"
ln -sfT ${PWD} "$P/$N-$x-$y.$T.enable/$T"
popd 1>/dev/null
}
function populateDirs () {
cd "$P" || exit
for i in $P/*.$T.enable ; do
[ -L "$i/$T" ] || continue
[ -L "$i" ] && continue
[ -f "$i" ] && continue
cd "$i" || continue
d="$(readlink -n -s -f $T )"
for j in $T/* ; do
[ -L "$j" ] && continue
[ -d "$j" ] && continue
j="$(basename $j)"
if [ "$j" = "parent" ] ; then
continue
elif [ "$j" = "ChangeLog" ] ; then
loc="$(basename $i)"
ln -s "$d/$j" "$i/$j.${loc/.$T.enable}"
else
ln -s "$d/$j" "$i/$j"
fi
done
done
}
function writeParentFile () {
[ -f "$P/parent" ] && rm "$P/parent"
cd "$P" || exit
ls -1 \
| egrep -e'enable$' \
| LC_COLLATE=C sort -g >> "$P/parent"
}
[ "$1" != "" ] && exit
[ -L "$P" ] && echo "$P\n must NOT be a link but instead:\n$P/$T" \
&& exit
[ -d "$P" ] || exit
[ -L "$P/$T" ] || exit
removeOldStackDirs
declare -i N=100
export N
Profile="$(readlink $P/$T)"
export PBase="${Profile%profiles*}profiles"
makeStackDirs $Profile
N=$(( $N - 100 ))
populateDirs
echo " --- writeParentFile"
writeParentFile
echo " generated: $(( $N / 10 ))"
|
Such a resulted profile you can easily manipulate and learn howto create your own release: Just add your own sub directories named *.enable (but not *.target.enable), holding your own mask and use files.
Last edited by ulenrich on Thu Jul 24, 2014 8:56 pm; edited 1 time in total |
|
| Back to top |
|
|
hasufell
Retired Dev
Joined: 29 Oct 2011
Posts: 429
|
| Posted: Thu Jul 24, 2014 8:53 pm Post subject: |
|
|
| ulenrich wrote: | | You can do it without any GLEPs |
Then it's simply unsupported hackery. |
|
| Back to top |
|
|
ulenrich
Veteran
Joined: 10 Oct 2010
Posts: 1480
|
| Posted: Thu Jul 24, 2014 8:58 pm Post subject: |
|
|
| hasufell wrote: | | ulenrich wrote: | | You can do it without any GLEPs |
Then it's simply unsupported hackery. |
For sure. But maybe there will be a feature request of more people than just me .... and
- many more participants making many more Gentoo releases
- some systemd discussions obsolete
- newbies will find what they think Gentoo is
- growth in adoption of Gentoo techniques
will be the result. |
|
| Back to top |
|
|
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
| Posted: Fri Jul 25, 2014 7:07 pm Post subject: |
|
|
| hasufell wrote: | | Most of what happens in overlays never ends up in the tree, because no one knows and no one communicates. Even projects like science that do have a relatively strong userbase contributing to their official overlay... rarely import that stuff into the tree. |
I don't see that as an issue, even if I would prefer it if their work made it to tree. If someone is interested in such specialist work, then they can use the overlay, and join the IRC channel for help and collaboration.
| Quote: | | Sunrise is dying too. Our collaboration channels have basically failed. Our system is not designed for it either, which is why we keep adding developers and make things worse. |
Sunrise always appeared to me as something most devs ignore, and at least initially I didn't want to add an overlay with random packages in. We can mask by overlay now, but it's still not something I'd want to add.
As for channels failing etc, Gentoo isn't a software house, and nor is the main product its own software, but rather bash scriptlets. It's hardly the same as the kernel, and there is much more scope for people to work in their own areas without needing to collaborate, since all they're doing is small bash scriptlets to wrap a build process. It's much more of an ecosystem than anything else.
| Quote: | | steveL wrote: | | Indeed: I'd argue you could say the same about FLOSS in general, and most of your points are really about that. |
Not all projects mess up as bad as gentoo. No, definitely not. |
Not sure what you're talking about now; "it works for a lot of people in a very weird way". As I said, Gentoo isn't a software project like most others; there isn't one product that's being worked on. The closest you get is portage. So you're comparing radically different types of project.
Where Gentoo does mess up imo, is in technical clarity. There's no willingness to rework stuff based on design, with the experience gained from previous development, likely because most devs are complete and utter beginners, in real-world terms.
Far too often you're beholden to one or two people who frankly have no real depth to them. But that's a sociopolitical thing, which devs try to pretend doesn't exist (until they want to beat up on a user, when suddenly everyone's very big on social rules to justify devs behaving like asshats.)
It's all par for the course from young males fresh outta Uni, without much real-world experience. Pack mentality, bless em.
It's funny I can remember a few threads like this over the years; one has to wonder why you can't have this discussion with your fellow developers, although the pack mentality is probably why. And ofc discussion isn't much use without action thereafter; usual pattern is someone says "wouldn't it be better if we ..", it gets discussed and pondered from all sorts of angles that have nothing to do with anything practical, no-one really argues; and nothing changes.
Meanwhile others who clearly haven't got a clue are stomping all over the tree and proud that they finally got something to sorta work after two years (really badly, and missing the central point) which could have been done much better over a few months in overlay. And if we say anything, we get the wagons circling and the pack mentality to deal with. So I guess I agree with you, though not for the same reasons. Gentoo's main problem is the attitude of the developers, since it turns away much more useful people, who got over that behaviour at 12.
It's the poison-people discussion all over again, which to me simply means you need stronger sociopolitical safeguards, not weaker ones. Or you get wacky ideas like what ulenrich has suggested being tried for real, or worse as unspoken agendas, and spews of totally inappropriate dependencies clogging up everyone's install for zero benefit, and lots of pain.
But ultimately, it's not your software. Which is why most of it still works; the people putting it out typically aren't as parochial, and concerned with only one OS. No serious developer would ever tie themselves to one platform like that.
Anyhow you seem to be on a downer and acting as if Gentoo will die, so I suggest a timeout. You'll feel much better after a couple of weeks off. |
|
| Back to top |
|
|
Anon-E-moose
Watchman
Joined: 23 May 2008
Posts: 6098
Location: Dallas area
|
| Posted: Fri Jul 25, 2014 7:57 pm Post subject: |
|
|
| steveL wrote: | | Anyhow you seem to be on a downer and acting as if Gentoo will die, so I suggest a timeout. You'll feel much better after a couple of weeks off. |
I can't say that I disagree with the gentoo will die assessment (it seems some devs are working on it).
Look at the couple of devs that "constantly" show up on sysd threads trying to derail any discussion.
I don't trust those devs any further than I can throw them as I sense another interest in their "contributions"
than what's good for gentoo. I think their allegiance is to another distro entirely.
But I understand where hasufell is coming from. And deep inside I think you do too, one only has to look back
at the whole separate /usr thing and the council vote about it. I'm already seeing things crop up in ebuilds that
I'm not happy about, and have been making local copies and undoing the "stupid" choices that are being
made in them, mostly, in my case, having to do with stupid version dependencies. As I find that using an older
version of packages (than the current available) works just fine. I don't know if it's deliberate, as the whole
gnome show has become or just due to devs that are too stupid/lazy to investigate further than "it works with the latest".
Anyway.... [/rant]
_________________
PRIME x570-pro, 3700x, 6.1 zen kernel
gcc 13, profile 17.0 (custom bare multilib), openrc, wayland |
|
| Back to top |
|
|
hasufell
Retired Dev
Joined: 29 Oct 2011
Posts: 429
|
| Posted: Fri Jul 25, 2014 9:07 pm Post subject: |
|
|
| steveL wrote: | | hasufell wrote: | | Most of what happens in overlays never ends up in the tree, because no one knows and no one communicates. Even projects like science that do have a relatively strong userbase contributing to their official overlay... rarely import that stuff into the tree. |
I don't see that as an issue, even if I would prefer it if their work made it to tree. If someone is interested in such specialist work, then they can use the overlay, and join the IRC channel for help and collaboration. |
Then you don't seem to understand the issue about why decentralized packaging in gentoo is broken.
| steveL wrote: | | As for channels failing etc, Gentoo isn't a software house, and nor is the main product its own software, but rather bash scriptlets. It's hardly the same as the kernel, and there is much more scope for people to work in their own areas without needing to collaborate, since all they're doing is small bash scriptlets to wrap a build process. It's much more of an ecosystem than anything else. |
I'm not sure why you explain gentoo to me. I'm working on it every day.
The core of gentoo is still the package manager and the specification behind it, because that defines how the scriptlets look like and how they interact with each other and the package manager.
So, I still don't see your point, except that the analogy wasn't perfect.
| steveL wrote: | | It's funny I can remember a few threads like this over the years; one has to wonder why you can't have this discussion with your fellow developers, although the pack mentality is probably why. And ofc discussion isn't much use without action thereafter; usual pattern is someone says "wouldn't it be better if we ..", it gets discussed and pondered from all sorts of angles that have nothing to do with anything practical, no-one really argues; and nothing changes. |
Interesting. Why do you think I did not have this discussion with my fellow developers? Wild guess?
For the rest, I can't figure out how that differs much from what I said, except that you seem to weigh your points differently
| hasufell wrote: | | If you want you could say that gentoo is broken technically (I skipped that large part about portage, eclasses and other things), organizationally and socially. |
|
|
| Back to top |
|
|
yagami
Apprentice
Joined: 12 May 2002
Posts: 269
Location: Leiria, Portugal
|
| Posted: Fri Jul 25, 2014 9:47 pm Post subject: |
|
|
| Anon-E-moose wrote: | | steveL wrote: | | Anyhow you seem to be on a downer and acting as if Gentoo will die, so I suggest a timeout. You'll feel much better after a couple of weeks off. |
I can't say that I disagree with the gentoo will die assessment (it seems some devs are working on it).
Look at the couple of devs that "constantly" show up on sysd threads trying to derail any discussion.
I don't trust those devs any further than I can throw them as I sense another interest in their "contributions"
than what's good for gentoo. I think their allegiance is to another distro entirely.
But I understand where hasufell is coming from. And deep inside I think you do too, one only has to look back
at the whole separate /usr thing and the council vote about it. I'm already seeing things crop up in ebuilds that
I'm not happy about, and have been making local copies and undoing the "stupid" choices that are being
made in them, mostly, in my case, having to do with stupid version dependencies. As I find that using an older
version of packages (than the current available) works just fine. I don't know if it's deliberate, as the whole
gnome show has become or just due to devs that are too stupid/lazy to investigate further than "it works with the latest".
Anyway.... [/rant] |
I know that although the conversation was good, it was all going to end in a s**t**d discuttion and insulting the s**t**d supporting devs.
I know that more and more I use Manjaro ( which by the way has an openRC edition ) and more and more leave Gentoo, but not by technical matters. maybe i am getting old, but i dont actually see any point or fun in these discutions and insults of who is the biggest geek. ( this rant is not specifically to you , anonymous moose, but to the topic of whats bad on gentoo community ).
probably in technical terms, gentoo is missing what funtoo has, or so to be, the "fun" and innovation. but ... gentoo is still working well, at least on my pc's |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
Gentoo Chat
