Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
[solved] Start new branch of Gentoo??
View unanswered posts
View posts from last 24 hours

Goto page 1, 2, 3  Next  
Reply to topic    Gentoo Forums Forum Index Gentoo Chat
View previous topic :: View next topic  
Author Message
derverstand
Guru
Guru


Joined: 15 Dec 2005
Posts: 511
Location: /dev/null

PostPosted: Sat Sep 23, 2006 9:04 pm    Post subject: [solved] Start new branch of Gentoo?? Reply with quote

Hi,

At the moment I really sick of the unprofessional release management of Gentoo. But as I think this is only a problem caused by the people, who should take care of the releases and not the portage / ebuild concept itself, I want to start an own portage branch. The concept was already posted by me in an other thread, but not discussed by the others. So I just do it on my own.

The ideas are:
(1) Complete mirrors (no third party dependencies for sources / patches any more)
(2) Different releases of portage:
(2a) Stable Gentoo (for example no useless updates to Xorg7, gcc-4 etc)
(2b) Unstable Gentoo (the current branch)
(2c) Experimental (tu current masked packages)
(3) More freedom to the user: Show changelog for every package to update, so everybody can decide weather or not the update is good.

At the moment I'm doing exactly this on my own server. My question is: Are there other people interested in this concept? If so, how to distribute the portage / complete mirrors?

Best regards!


Last edited by derverstand on Thu Sep 28, 2006 11:50 am; edited 1 time in total
Back to top
View user's profile Send private message
Cinquero
Apprentice
Apprentice


Joined: 24 Jun 2004
Posts: 249

PostPosted: Sat Sep 23, 2006 9:46 pm    Post subject: Reply with quote

deleted

Last edited by Cinquero on Fri Sep 29, 2006 8:40 pm; edited 1 time in total
Back to top
View user's profile Send private message
Eduard Munteanu
n00b
n00b


Joined: 06 Sep 2006
Posts: 20
Location: Bucuresti, Romania

PostPosted: Sat Sep 23, 2006 10:36 pm    Post subject: Reply with quote

Cinquero>
1) That's why ebuilds don't get modified, but revisions are released ("-r"s). Older and broken ebuilds can be masked/removed from the portage tree.
3) I don't want to be rude, but the idea sucks. Imagine the developers sorting through hundreds of bogus bug reports because someone broke it's filesystem or did something he wasn't supposed to do.

Making the process of filing a bug report easier is a good idea, but automating it blindly is evil.

derverstand>
1) The reasons you sometimes need to download packages manually are legal problems. Not all the packages inside the portage tree are free software / opensource. No fork can help solve this matter.
2) I honestly don't understand what you mean.
3) Well, you must have a good reason not to update, especially when running the stable tree.
Back to top
View user's profile Send private message
Cinquero
Apprentice
Apprentice


Joined: 24 Jun 2004
Posts: 249

PostPosted: Sat Sep 23, 2006 10:51 pm    Post subject: Reply with quote

deleted

Last edited by Cinquero on Fri Sep 29, 2006 8:41 pm; edited 1 time in total
Back to top
View user's profile Send private message
drwook
Veteran
Veteran


Joined: 30 Mar 2005
Posts: 1322
Location: London

PostPosted: Sat Sep 23, 2006 11:06 pm    Post subject: Re: Start new branch of Gentoo?? Reply with quote

derverstand wrote:

The ideas are:
(1) Complete mirrors (no third party dependencies for sources / patches any more)
(2) Different releases of portage:
(2a) Stable Gentoo (for example no useless updates to Xorg7, gcc-4 etc)
(2b) Unstable Gentoo (the current branch)
(2c) Experimental (tu current masked packages)
(3) More freedom to the user: Show changelog for every package to update, so everybody can decide weather or not the update is good.


1/
i/ Legal/licencing issues around e.g. sun-jdk, ibm-jdk? Am sure there are plenty more but just some examples I can think of off the top of my head
ii/ Increases hosting costs mirroring more files (yeah, not the biggest concern perhaps but should be mentioned)
2/ Explain. There are different releases if you mean the incremental versions, otherwise I don't have a clue what you're on about, unless it's paludis?
2a/ Why not an 'uber stable' branch without useless updates to gcc-3, 2.4/2.6 kernel, etc? Gentoo does not force anyone to update...
2b/ You mean the current stable branch? It's pretty rock solid IMHO...
2c/ All of them? Just call it the chaos branch....
3/ umm,
Code:
cat /usr/portage/{category}/{package}/ChangeLog
you mean?

Why waste so much energy trying to solve problems that don't exist?
Back to top
View user's profile Send private message
Eduard Munteanu
n00b
n00b


Joined: 06 Sep 2006
Posts: 20
Location: Bucuresti, Romania

PostPosted: Sat Sep 23, 2006 11:16 pm    Post subject: Reply with quote

Cinquero wrote:
Now tell me: how do you mark a stable ebuild as unstable? By declaring a newer release as unstable? Won't work. emerge will use the older release which is broken. This is why the current system is inherently broken.

Just mask it. A simple emerge -auv world will automatically fix the problem (by downgrading/upgrading that package). Everybody should do this once in a while.

Cinquero wrote:
1.) Did I say "devs"? No.
2.) If the issue is real, many bug reports are similar if not identical. So what?

Unless there's an automated system to accurately classify bugs by their similarity at the other end (devs) too, it's still going to be a huge problem. And simply because there are lots of similar bugs is not a way to reliably get bugs on a worklist. This system is going to replace "hey, there are not enough bug reports" with "damn, I have to go through [insert a big number here] reports to find a real bug", which I think it's worse.

I personally believe the forums are a great filter for bogus bug reports.

Cinquero wrote:
I did not say that the reports should be filed with bugzilla. I did not even say how to evaluate them. I just said it would possibly be a quite useful resource to detect problems immediately.

Thanks for being negative on something not even discussed.

I didn't want it to sound like that :(, but it seemed almost impossible to do it well. About having a separate & automated bug database, it could be a good idea, but I doubt any devs would take a look through it. I agree it could be useful in case some user wants to rule out his error, by looking through the auto bug db for symptoms he's experiencing, but nothing more. And this automated system should warn the user not to quit asking on forums or filing manual bug reports once he's sure that's a real problem.
Back to top
View user's profile Send private message
Voltago
Advocate
Advocate


Joined: 02 Sep 2003
Posts: 2592
Location: userland

PostPosted: Sat Sep 23, 2006 11:41 pm    Post subject: Reply with quote

A less invasive idea to reduce updates:
A new standard ebuild variable 'COMPELLING'.

It works like this:
If you run
Quote:
emerge -uD world

an update of a stable (arch) package will only be performed, if there is at least one higher versioned stable package with the 'COMPELLING' property set. In that case, the highest stable package version is chosen.
For unmasked unstable (~arch) packages, everthing stays the same.

The 'COMPELLING' flag is only to be set, if
1) there is considerable new functionality in a higher version
2) there are severe and/or security relevant bugs fixed in the new package

Of course, there will be a command line option that restores the default behavior of emerge, and rest assured, there will be plenty of people who use it all the time and complain about all the unnecessary updates.
A similar mechanism can be employed for libraries that introduce ABI breakages.


Last edited by Voltago on Sun Sep 24, 2006 12:00 am; edited 2 times in total
Back to top
View user's profile Send private message
Cinquero
Apprentice
Apprentice


Joined: 24 Jun 2004
Posts: 249

PostPosted: Sat Sep 23, 2006 11:52 pm    Post subject: Reply with quote

deleted

Last edited by Cinquero on Fri Sep 29, 2006 8:41 pm; edited 1 time in total
Back to top
View user's profile Send private message
mark_alec
Bodhisattva
Bodhisattva


Joined: 11 Sep 2004
Posts: 6066
Location: Melbourne, Australia

PostPosted: Sun Sep 24, 2006 1:51 am    Post subject: Reply with quote

Moved from Portage & Programming to Gentoo Chat.
_________________
www.gentoo.org.au || #gentoo-au
Back to top
View user's profile Send private message
Archangel1
Veteran
Veteran


Joined: 21 Apr 2004
Posts: 1212
Location: Work

PostPosted: Sun Sep 24, 2006 4:42 am    Post subject: Reply with quote

Voltago wrote:
A less invasive idea to reduce updates:
A new standard ebuild variable 'COMPELLING'.

It works like this:
If you run
Quote:
emerge -uD world

an update of a stable (arch) package will only be performed, if there is at least one higher versioned stable package with the 'COMPELLING' property set. In that case, the highest stable package version is chosen.
For unmasked unstable (~arch) packages, everthing stays the same.

The 'COMPELLING' flag is only to be set, if
1) there is considerable new functionality in a higher version
2) there are severe and/or security relevant bugs fixed in the new package

Of course, there will be a command line option that restores the default behavior of emerge, and rest assured, there will be plenty of people who use it all the time and complain about all the unnecessary updates.
A similar mechanism can be employed for libraries that introduce ABI breakages.

But you'd end up doing every update after the first anyway...

Say you've got foo-1.0 installed. Then foo-2.0 is released, with "COMPELLING" set, so Portage upgrades you. All sweet.
Now foo-2.0-r1 is released, which also has "COMPELLING" set because it supersedes 2.0, so anyone still on 1.0 should upgrade to 2.0-r1. But anyone on 2.0 will be upgraded too. This continues through all updates to foo in future, because they're all compelling compared to 1.0.
_________________
What are you, stupid?
Back to top
View user's profile Send private message
steveL
Watchman
Watchman


Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery

PostPosted: Sun Sep 24, 2006 5:14 am    Post subject: Reply with quote

Cinquero wrote:
Eduard Munteanu wrote:

Just mask it. A simple emerge -auv world will automatically fix the problem (by downgrading/upgrading that package). Everybody should do this once in a while.


I consider that an upstream issue -- not an user issue. That way of handling such things is pretty, uhm, sub-optimal.

Is there really no way to change a package from stable to unstable?
Quote:

Quote:
Unless there's an automated system to accurately classify bugs by their similarity at the other end (devs) too, it's still going to be a huge problem. ...


Agreed.

Quote:

I didn't want it to sound like that :(, but it seemed almost impossible to do it well. About having a separate & automated bug database, it could be a good idea, but I doubt any devs would take a look through it. I agree it could be useful in case some user wants to rule out his error, by looking through the auto bug db for symptoms he's experiencing, but nothing more. And this automated system should warn the user not to quit asking on forums or filing manual bug reports once he's sure that's a real problem.


That usefulness for the user(!) is exactly what should be worked on in any program.
Further more it doesn't have to go in the main bugzilla.
The interface could also automatically look up the relevant package in bugzilla.
Quote:

The actual point is to attach solutions to the bug reports -- possibly provided by the users themselves (in the spirit "that worked for me"). The crux is that devs could then possibly have a look at those solutions -- at least at those known to be successful. In the end, the devs wouldn't be involved at all. It would be a totally independent and additional stage to provide fixes and solutions faster and more tightly integrated with the emerge process, some sort of a formalized user2user and per ebuild communication channel *g*.

Great idea!
Back to top
View user's profile Send private message
pjp
Administrator
Administrator


Joined: 16 Apr 2002
Posts: 18551

PostPosted: Sun Sep 24, 2006 5:16 am    Post subject: Reply with quote

There is a GLEP for a "stable" portage tree... 19 I think.
_________________
Your lips move, but I can't hear what you're saying.
Back to top
View user's profile Send private message
drwook
Veteran
Veteran


Joined: 30 Mar 2005
Posts: 1322
Location: London

PostPosted: Sun Sep 24, 2006 8:15 am    Post subject: Reply with quote

Maybe I should look into writing a quick script to populate package.mask with any ebuilds within the first 5 years in the tree (and remove anything now >5 years old as long as it's marked stable)? Only problem being I doubt much if anything would survive in the tree that long, but thinking maybe not that crazy an idea if user-definable timespan (e.g. "I don't want anything new until it's been stable for x months") or hold 'old' ebuilds that have been cleaned off the tree in a SVN overlay or something...

I'd guess something similar was the inspiration of Debian's stable, which seems to be what people are asking for :twisted:
Back to top
View user's profile Send private message
steveL
Watchman
Watchman


Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery

PostPosted: Sun Sep 24, 2006 10:02 am    Post subject: Reply with quote

The real problem I can see is that people want to be able to install a particular distro and keep getting updates for it.

This doesn't just mean being able to upgrade to the latest software; it means that changes won't happen in a stable setup that break a system in production use. To my mind this is what holds gentoo back against other distros.

Thus no major upgrades to GCC, which officially requires rebuilding the whole system and world trees, or Xorg which seems to have tripped up people on the change from 6 to 7. Gentoo 1.4 was another difficult change system-wide IIRC.

If at a later date the user wants to make the plunge, that's fine (after all lots of other people will have updated by then.)

I guess what it boils down is being able to hold off on any major changes, whilst keeping up with other (maintenance, security + apps) software updates.

This could be done by classing certain packages as critical (after all it's usually core packages that we're talking about) /if installed/. Then standard emerge system or world would not automatically upgrade those beyond a certain spec (eg GCC M.x.y when M changes) . The admin hassle might come from having later stable packages depending on updated software, but that's what slots are for, and eventually the user will update when their fave software won't upgrade to the shiniest version.
Back to top
View user's profile Send private message
Eduard Munteanu
n00b
n00b


Joined: 06 Sep 2006
Posts: 20
Location: Bucuresti, Romania

PostPosted: Sun Sep 24, 2006 10:59 am    Post subject: Reply with quote

It's kind of stupid not to update to a new minor release, if that's what you're talking about. It sounds like "I'm installing Redhat 9 because it's old&stable / the machine is too old / always worked for me". Everybody should use either the latest version or the latest revision of an older but supported major release (e.g: GCC 3.4.6 and 4.1.1 are a good idea, but 2.95 I think not).

OTOH, if you mean something like not upgrading from GCC 3.4 to 4.1, then it's a legitimate reason. That's what profiles are for. Why not use them? An older but supported major release problem is easily solved if you select an older profile (but not one 5 years old, I think it's going to hurt a lot more than help).

What's the problem with emerge -auv world? Is it updating all your packages and you don't have time for this? No problem, create a meta ebuild and run emerge on that.

steveL wrote:
Is there really no way to change a package from stable to unstable?

This issue is more theoretical than practical. Usually, packages which go stable are meant to stay there. I'm asking again, what's the problem with masking?

So, please enlighten me: why this custom script hassle?!
Back to top
View user's profile Send private message
derverstand
Guru
Guru


Joined: 15 Dec 2005
Posts: 511
Location: /dev/null

PostPosted: Sun Sep 24, 2006 11:01 am    Post subject: Reply with quote

-> What I mean by releases: Take for example to Xorg 6.9 It was working fine. Since the update to Xorg >= 7 none of my machines has DRI any more. So why is my system broken, if I keep the working 6.9?? And why is my system broken, if I use gcc <4?? This is exactly the point: I just want to know, what is going on in the new releases (-> changelog). I don't need no colorized output-"fixes" with graven feet...
-> Automerging & Bugreporting: Good idea. But I want to be independent from any webinterface. Bugs should be automatically (maybe option for asking) posted into a bugzilla. Keep the spirit of portage: shellscript!
-> Seperate stability-file: Definitively no. Why should one make it more complex? The full functionality is already included in portage. It only should be utilized more. And if (I could understand this) it is to much effort to check every new ebuild completely, then one should add an autobugreport-system!
-> Mirrors: Of course I'm fully aware of the license / legal problems. But there should be a script to build a local mirror on your own (I have one, which is far from good ;). As I'm running a couple of PCs (some of them in a diskless MIPS-burning cluster), I need an own mirror (>15 PCs updated via DSL??). And this is propably the case for more people...
-> "Minor updates" Some of them (like one of the last base-layouts) are breaking configuration files (in this case: /etc/conf.d/net -> routing entries). So one has to be very careful, when updating to "minor" releases, too.

Best regards.
Back to top
View user's profile Send private message
Eduard Munteanu
n00b
n00b


Joined: 06 Sep 2006
Posts: 20
Location: Bucuresti, Romania

PostPosted: Sun Sep 24, 2006 11:52 am    Post subject: Reply with quote

derverstand wrote:
-> "Minor updates" Some of them (like one of the last base-layouts) are breaking configuration files (in this case: /etc/conf.d/net -> routing entries). So one has to be very careful, when updating to "minor" releases, too.

If you use -5 when doing etc-update, then I must say this is your fault. Please stop complaining if you don't use these tools as you should.

derverstand wrote:
-> Mirrors: Of course I'm fully aware of the license / legal problems. But there should be a script to build a local mirror on your own (I have one, which is far from good ;). As I'm running a couple of PCs (some of them in a diskless MIPS-burning cluster), I need an own mirror (>15 PCs updated via DSL??). And this is propably the case for more people...

And how is this going to be implemented by a Gentoo fork?

Well, I agree older profiles should be freezed to specific major releases. I don't really know whether xorg has a problem with version capping in profiles previous to 2006.1. OTOH, changing your profile has some consequences you have to deal with (like this upgrade from xorg 6.9 to 7).
Back to top
View user's profile Send private message
Voltago
Advocate
Advocate


Joined: 02 Sep 2003
Posts: 2592
Location: userland

PostPosted: Sun Sep 24, 2006 1:45 pm    Post subject: Reply with quote

Archangel1 wrote:
But you'd end up doing every update after the first anyway...

Say you've got foo-1.0 installed. Then foo-2.0 is released, with "COMPELLING" set, so Portage upgrades you. All sweet.
Now foo-2.0-r1 is released, which also has "COMPELLING" set because it supersedes 2.0, so anyone still on 1.0 should upgrade to 2.0-r1. But anyone on 2.0 will be upgraded too. This continues through all updates to foo in future, because they're all compelling compared to 1.0.

You misunderstood me: Maybe foo-2.0-r1 is only a fix to a minor problem that mucks up installation for a limited number of people. So it isn't necessarily marked COMPELLING. Now everyone on 1.0 will still be upgraded to 2.0-r1, because there is one higher stable COMPELLING version; it doesn't matter which one that is. The 2.0 users won't be upgraded because there is no higher stable COMPELLING version. But as soon as every little bugfix is marked COMPELLING, then I agree, the whole thing is useless.


Last edited by Voltago on Sun Sep 24, 2006 4:17 pm; edited 3 times in total
Back to top
View user's profile Send private message
derverstand
Guru
Guru


Joined: 15 Dec 2005
Posts: 511
Location: /dev/null

PostPosted: Sun Sep 24, 2006 3:08 pm    Post subject: Reply with quote

Okay. I think this makes no sense.... I DO USE THE TOOLS. And don't have time to waste for trivial discussions. If nobody is interested: ok. My gentoo release is working.

Best regards.
Back to top
View user's profile Send private message
Voltago
Advocate
Advocate


Joined: 02 Sep 2003
Posts: 2592
Location: userland

PostPosted: Sun Sep 24, 2006 3:24 pm    Post subject: Reply with quote

derverstand wrote:
Okay. I think this makes no sense.... I DO USE THE TOOLS. And don't have time to waste for trivial discussions. If nobody is interested: ok. My gentoo release is working.

Well, your mind is quite made up it seems, and I do by no means intend to discourage you, but, realistically, you have to ask yourself the following questions:
Are you aware what amount of work will go into maintaining a portage tree fork? Try to make a realistic estimate on the man-power you need. Who will help you? How long will you be able to keep up that effort?
Of course, now you start off with the best intentions, and with investing a considerable amount of your life-time you will be able to keep your project going for a while. But when the novelty wears off and every-day tasks seem duller and duller each time and people don't seem to give any credit to your labouring but start bitching instead about slow service, will you still go on?
If you kind of hope that Gentoo users with the skills for maintainance and QA and time to spare will simply flock to your banner once you got your project going, then wake up: This is not going to happen.

And I realize that my initial statement was a lie. I want to discourage you. Rather than starting a partisan tree, help out with QA under the hood of the regular project. That way, your time and effort will not be wasted when you decide to stop.
Back to top
View user's profile Send private message
Eduard Munteanu
n00b
n00b


Joined: 06 Sep 2006
Posts: 20
Location: Bucuresti, Romania

PostPosted: Sun Sep 24, 2006 4:59 pm    Post subject: Reply with quote

derverstand wrote:
Okay. I think this makes no sense.... I DO USE THE TOOLS. And don't have time to waste for trivial discussions. If nobody is interested: ok. My gentoo release is working.

Best regards.

You misunderstood what I said. It was not about using etc-update or not. It was about using etc-update the right way, by inspecting the diffs carefully ("non-trivial changes" as etc-update names them). It's your fault if it overwrites an important file in /etc.

I might sound too negative regarding this matter, but I have good reasons to do so. I don't like hearing complaints about Gentoo management, like
derverstand wrote:
At the moment I really sick of the unprofessional release management of Gentoo

, when in fact the user is guilty or has no idea what he's talking about. And I'm yet to hear professional (as opposed to "unprofessional release management of Gentoo" :twisted: ) solutions and not this nonsense I've been reading in the other posts of this thread.

I mean no offense to people who posted realistic ideas/comments.
Back to top
View user's profile Send private message
Headrush
Watchman
Watchman


Joined: 06 Nov 2003
Posts: 5597
Location: Bizarro World

PostPosted: Sun Sep 24, 2006 5:37 pm    Post subject: Reply with quote

Voltago wrote:
And I realize that my initial statement was a lie. I want to discourage you. Rather than starting a partisan tree, help out with QA under the hood of the regular project. That way, your time and effort will not be wasted when you decide to stop.

Unless you've been involved in a project of any size, I believe most people underestimate the time and effort actually involved in maintaining any type of project.
Except in the cases where a fundamental ideological change is involved or needed, I believe in most cases you get better return on resources when improving the current system. (If possible.)
Back to top
View user's profile Send private message
renrutal
Tux's lil' helper
Tux's lil' helper


Joined: 26 Mar 2005
Posts: 135
Location: Brazil

PostPosted: Sun Sep 24, 2006 10:30 pm    Post subject: Reply with quote

I believe derverstand is just using the wrong distro, or at least he wants a [binary] distribution, not a meta-distribution. Each distro is build for one or more purposes, there are the ones specialized in desktop, others in content creation, servers, etc. Gentoo is specialzed in making distributions, or customized builds of a system.

Seeing the examples you gave, it appears to be more a problem between keyboard and chair than something concrete.

I do agree with some of the problems, I've seen many of them in those 3 years using Gentoo, both in stable and testing. However I also realise it's impossible to provide a solution for all X number of different possible system builds. It's a problem binary distros don't have, but we do try.
Back to top
View user's profile Send private message
drwook
Veteran
Veteran


Joined: 30 Mar 2005
Posts: 1322
Location: London

PostPosted: Mon Sep 25, 2006 7:40 am    Post subject: Reply with quote

Quite. If you want debian stable, use debian stable by all means...
Back to top
View user's profile Send private message
tsunam
Retired Dev
Retired Dev


Joined: 23 Feb 2004
Posts: 343

PostPosted: Mon Sep 25, 2006 5:37 pm    Post subject: Re: Start new branch of Gentoo?? Reply with quote

derverstand wrote:
Hi,

At the moment I really sick of the unprofessional release management of Gentoo. But as I think this is only a problem caused by the people, who should take care of the releases and not the portage / ebuild concept itself, I want to start an own portage branch. The concept was already posted by me in an other thread, but not discussed by the others. So I just do it on my own.

The ideas are:
(1) Complete mirrors (no third party dependencies for sources / patches any more)


Have fun violating licenses then. I'm sure you won't get letters to cease what you are talking about doing.

Quote:
(2) Different releases of portage:
(2a) Stable Gentoo (for example no useless updates to Xorg7, gcc-4 etc)


I'm so glad that both xorg7 and gcc-4 are useless updates. As upstream for both consider xorg 6.8 and gcc less then 4 to be dead and won't support it any longer. If you consider them worthless though...more power to you. If you wish you could go use OpenBSD, the upcoming release has Gcc 2.95.3 (+ patches) and 3.3.5 (+ patches).
_________________
I'm not afraid of happy endings, just afraid my life wont work that way.
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Gentoo Chat All times are GMT
Goto page 1, 2, 3  Next
Page 1 of 3

 
Jump to:  
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