[aur-general] [community] repository cleanup
Hi TUs, we are here again. After the success with [extra] (DEVs adopted ~80 packages, TUs ~60), I want to reduce the number of orphans packages in [community]. Actually, they are 84 (82, I just adopted two...). The list is here[1]. Simply cross out the package which you want to maintain in [community]. Adoption is not required, but would be nice. Packages will be moved to AUR this Saturday 20th. Thanks [1] https://wiki.archlinux.org/index.php/DeveloperWiki:Community_Repo_Cleanup -- Andrea Scarpino Arch Linux Developer
Am Tue, 16 Nov 2010 11:34:11 +0100 schrieb Andrea Scarpino <andrea@archlinux.org>:
Hi TUs, we are here again. After the success with [extra] (DEVs adopted ~80 packages, TUs ~60), I want to reduce the number of orphans packages in [community]. Actually, they are 84 (82, I just adopted two...).
The list is here[1]. Simply cross out the package which you want to maintain in [community]. Adoption is not required, but would be nice. Packages will be moved to AUR this Saturday 20th.
Thanks
[1] https://wiki.archlinux.org/index.php/DeveloperWiki:Community_Repo_Cleanup
squashfs-tools What have you told me about that when you have moved this from [extra] to [community]? Now it shall be moved to AUR? And why was it moved to [community] if it is an orphan there? And isn't this needed to building LiveCDs including the Arch Linux installation CDs? You really should think about that. Heiko
On Tuesday 16 November 2010 12:03:14 Heiko Baums wrote:
squashfs-tools
What have you told me about that when you have moved this from [extra] to [community]?
Now it shall be moved to AUR? And why was it moved to [community] if it is an orphan there? And isn't this needed to building LiveCDs including the Arch Linux installation CDs? Man you know that is a draft. I don't remember who wants to adopt squashfs- tools now, but surely a TU want because it has been moved to [community] recently.
-- Andrea Scarpino Arch Linux Developer
Am Tue, 16 Nov 2010 12:12:23 +0100 schrieb Andrea Scarpino <andrea@archlinux.org>:
Man you know that is a draft. I don't remember who wants to adopt squashfs- tools now, but surely a TU want because it has been moved to [community] recently.
Man, you know that you have removed most of the packages from your "draft" from [extra]. And I need to install three times as many packages from AUR as I needed three years ago just because they have been removed from the repos to AUR. And the repos of Arch Linux are pretty small compared to other distros anyway. That's always mentioned as disadvantage in reviews of Arch Linux. This is a bit compensated by AUR, but if you regularly move important or popular packages to AUR then Arch Linux will become a second Gentoo and for some people who want or need a binary distro unusable. And no, Arch Linux is not a distro from developers for developers anymore. This may have been in the beginning or Arch Linux but is not true anymore. Heiko
On 16/11/10 21:48, Heiko Baums wrote:
And no, Arch Linux is not a distro from developers for developers anymore. This may have been in the beginning or Arch Linux but is not true anymore.
It is true as long as the developers say it is true. Just because I tell GM their car is an aeroplane does not mean they will start flying. Allan
Am Tue, 16 Nov 2010 21:54:23 +1000 schrieb Allan McRae <allan@archlinux.org>:
It is true as long as the developers say it is true. Just because I tell GM their car is an aeroplane does not mean they will start flying.
Wrong. It is true as soon as a distro becomes more and more popular. And this is the case for Arch Linux. As long as a distro is unknown and only used by a few people, mainly its developers, this distro may only be from the devs for the devs. But as soon as it is mentioned together with and equivalent to the other big distros and gets more popular this is not true anymore independent from what a single developer says. Otherwise you should write a big note on the homepage and/or the download page that this distro is free but not meant to be used for the public. "Feel free to use it but don't expect anything. This distro is only meant to be used by us developers." Or something like this. Maybe a bit exaggerated. Heiko
On 16/11/10 22:06, Heiko Baums wrote:
Am Tue, 16 Nov 2010 21:54:23 +1000 schrieb Allan McRae<allan@archlinux.org>:
It is true as long as the developers say it is true. Just because I tell GM their car is an aeroplane does not mean they will start flying.
Wrong.
Really, cars will start flying if we just tell the manufactures it is so? Because I have been waiting for that for a long time...
It is true as soon as a distro becomes more and more popular. And this is the case for Arch Linux. As long as a distro is unknown and only used by a few people, mainly its developers, this distro may only be from the devs for the devs. But as soon as it is mentioned together with and equivalent to the other big distros and gets more popular this is not true anymore independent from what a single developer says.
But what actually makes you think that you can expect anything from the distro apart from whatever the developers decide? Just because people use the distribution does not mean their opinions count for anything. In fact, the opinions of >99% of the users of this distro do not count for anything because they contribute nothing towards the distribution. Now, if someone wants to pay me and others to develop the distro, then their opinions will count. Otherwise, I am "working" for free and so I am king of what I decide to do.
Otherwise you should write a big note on the homepage and/or the download page that this distro is free but not meant to be used for the public. "Feel free to use it but don't expect anything. This distro is only meant to be used by us developers." Or something like this. Maybe a bit exaggerated.
Slightly exaggerated. But perhaps we do need that to keep people from forming such unfounded ideas about their opinions counting towards anything around here. Allan
Am Tue, 16 Nov 2010 22:25:34 +1000 schrieb Allan McRae <allan@archlinux.org>:
Really, cars will start flying if we just tell the manufactures it is so? Because I have been waiting for that for a long time...
Do you always compare apples and oranges?
But what actually makes you think that you can expect anything from the distro apart from whatever the developers decide?
I generally don't expect anything. And I really respect the voluntary work. And I guess you haven't read many/any overbearing claims from me. But I expect from a developer that he doesn't remove popular packages inconsiderately from the repos. (Otherwise don't call Arch Linux a binary distro.) I expect from a developer that he uses his brain. Sorry for this wording. And I expect from a developer that he isn't arrogant and ignorant to users and that he listens to the user's opinion and respects their opinions.
Just because people use the distribution does not mean their opinions count for anything. In fact, the opinions of >99% of the users of this distro do not count for anything because they contribute nothing towards the distribution.
Have you thought of people who are not able to program or to write documentations or the like? Have you thought of people who just use their computers as an electronic typewriter or for looking for e-mails etc.? Do you really expect every usual computer user to learn a programming language? Btw., I'm already maintaining some AUR packages, worked on a wiki page for one of these packages. So don't tell me I wouldn't contribute anything.
Now, if someone wants to pay me and others to develop the distro, then their opinions will count. Otherwise, I am "working" for free and so I am king of what I decide to do.
You may decide what to do. But you should take the user's opinions or wishes into account.
Slightly exaggerated. But perhaps we do need that to keep people from forming such unfounded ideas about their opinions counting towards anything around here.
Really, these are not unfounded ideas. And you and some other devs (thankfully not all of them) should reflect about your ignorance. Heiko
On 16/11/10 23:09, Heiko Baums wrote:
And you and some other devs (thankfully not all of them) should reflect about your ignorance.
There is a difference between ignorance and not giving a shit... You seem ignorant to that difference. Allan
Am Wed, 17 Nov 2010 00:34:51 +1000 schrieb Allan McRae <allan@archlinux.org>:
There is a difference between ignorance and not giving a shit...
Right, giving a shit is also arrogant. Or where are the differences in your oppinion? Heiko
On 17/11/10 00:53, Heiko Baums wrote:
Am Wed, 17 Nov 2010 00:34:51 +1000 schrieb Allan McRae<allan@archlinux.org>:
There is a difference between ignorance and not giving a shit...
Right, giving a shit is also arrogant.
So now you are equating arrogant and ignorant? And did you miss a "not" there? Because "giving a shit" is rarely arrogant unless you take it too an extreme. In conclusion, you make very poor arguments with logical fallacies everywhere. No wonder you never seem to get your way. Allan (the ever pedantic bastard...)
On 11/16/2010 05:07 PM, Allan McRae wrote:
On 17/11/10 00:53, Heiko Baums wrote:
Am Wed, 17 Nov 2010 00:34:51 +1000 schrieb Allan McRae<allan@archlinux.org>:
There is a difference between ignorance and not giving a shit...
Right, giving a shit is also arrogant.
So now you are equating arrogant and ignorant? And did you miss a "not" there? Because "giving a shit" is rarely arrogant unless you take it too an extreme.
In conclusion, you make very poor arguments with logical fallacies everywhere. No wonder you never seem to get your way.
Allan (the ever pedantic bastard...)
GET A ROOM -- Ionuț
On Tuesday 16 November 2010 15:06:33 Ionuț Bîru wrote:
On 11/16/2010 05:07 PM, Allan McRae wrote:
On 17/11/10 00:53, Heiko Baums wrote:
Am Wed, 17 Nov 2010 00:34:51 +1000
schrieb Allan McRae<allan@archlinux.org>:
There is a difference between ignorance and not giving a shit...
Right, giving a shit is also arrogant.
So now you are equating arrogant and ignorant? And did you miss a "not" there? Because "giving a shit" is rarely arrogant unless you take it too an extreme.
In conclusion, you make very poor arguments with logical fallacies everywhere. No wonder you never seem to get your way.
Allan (the ever pedantic bastard...)
GET A ROOM
Awesome. Second only to Xyne's "Cared thou not, thou would have abstained" message:
On Tuesday 16 November 2010 15:06:33 Ionuț Bîru wrote:
On 11/16/2010 05:07 PM, Allan McRae wrote:
On 17/11/10 00:53, Heiko Baums wrote:
Am Wed, 17 Nov 2010 00:34:51 +1000
schrieb Allan McRae<allan@archlinux.org>:
There is a difference between ignorance and not giving a shit...
Right, giving a shit is also arrogant.
So now you are equating arrogant and ignorant? And did you miss a "not" there? Because "giving a shit" is rarely arrogant unless you take it too an extreme.
In conclusion, you make very poor arguments with logical fallacies everywhere. No wonder you never seem to get your way.
Allan (the ever pedantic bastard...)
GET A ROOM
Awesome. Second only in in terms of mailing list awesomeness to Xyne's "Cared thou not, thou would have abstained" message: http://mailman.archlinux.org/pipermail/aur-general/2010-September/010709.htm...
Am Tue, 16 Nov 2010 14:09:35 +0100 schrieb Heiko Baums <lists@baums-on-web.de>:
But I expect from a developer that he doesn't remove popular packages inconsiderately from the repos. (Otherwise don't call Arch Linux a binary distro.) I expect from a developer that he uses his brain. Sorry for this wording.
The Devs provide a basic distro + _some_ binary packages. We had a conclusion to provide a "core" repo that is known to be working and could be the ground to build _yourself_ a system to fit your needs. Take the "extra" repo as an additional gift some of us like to provide. It costs us a lot of time to maintain binary packages. Nobody is responsible to keep them alive. Packages come in and will fade out once a maintainer looses the interest in it. After some time we can't make sure it's still in a good shape and drop it from our official repo and leave it up to the community. You will have to live with this. -Andy
On Tue, 2010-11-16 at 22:25 +1000, Allan McRae wrote:
On 16/11/10 22:06, Heiko Baums wrote:
Otherwise you should write a big note on the homepage and/or the download page that this distro is free but not meant to be used for the public. "Feel free to use it but don't expect anything. This distro is only meant to be used by us developers." Or something like this. Maybe a bit exaggerated.
Slightly exaggerated. But perhaps we do need that to keep people from forming such unfounded ideas about their opinions counting towards anything around here.
Allan
I thought this idea of Heiko's was a great idea, myself. In a fit of inspiration, here's a suggestion. This would be placed after the last sentence of the 2nd paragraph in the 3-paragraph description. "This community involvement allows developers to focus on what's most important to them in the knowledge that the community will provide high quality packages for everything else." Just a draft, doesn't really express everything yet. Should I open a bug-report? And Heiko, I'm not sure why you're making a big fuss over these packages. Its not even much harder to work from the AUR. Comparisons to Gentoo are ridiculous. I've got 75 packages from the AUR currently, and I don't spend hours a day compiling. The longest package to compile is probably samba4 at 20 minutes, followed by the kernel at 10 minutes (optimized), everything else tops out at 1-2 minutes. Go have a cup of coffee, and consider the difference to the hours and hours you'd have spent compiling world on Gentoo.
Am Tue, 16 Nov 2010 21:27:50 +0800 schrieb Ng Oon-Ee <ngoonee@gmail.com>:
I thought this idea of Heiko's was a great idea, myself. In a fit of inspiration, here's a suggestion. This would be placed after the last sentence of the 2nd paragraph in the 3-paragraph description.
"This community involvement allows developers to focus on what's most important to them in the knowledge that the community will provide high quality packages for everything else."
Maybe there should be added that Arch Linux is only a distro from its devs for its devs and not from the community for the community. I wish you wouldn't add such a paragraph. Instead the ignorance of some devs should be removed. That was the better way.
And Heiko, I'm not sure why you're making a big fuss over these packages. Its not even much harder to work from the AUR. Comparisons to Gentoo are ridiculous. I've got 75 packages from the AUR currently, and I don't spend hours a day compiling. The longest package to compile is probably samba4 at 20 minutes, followed by the kernel at 10 minutes (optimized), everything else tops out at 1-2 minutes. Go have a cup of coffee, and consider the difference to the hours and hours you'd have spent compiling world on Gentoo.
I already used Gentoo for 6 years. So I know what that compiling means. That's why I was looking for a binary distro in the style of Gentoo. That's why I came to Arch Linux 3 or 4 years ago. And the more small packages need to be compiled from AUR the more time it takes compiling them. Many a little makes a mickle. My main problems are this mass cleanup and the ignorant responses I get from some devs to such remarks or to some bug reports etc. That's why I make such a fuss. Heiko
Excerpts from Heiko Baums's message of 2010-11-16 15:02:19 +0100:
Am Tue, 16 Nov 2010 21:27:50 +0800 schrieb Ng Oon-Ee <ngoonee@gmail.com>:
I thought this idea of Heiko's was a great idea, myself. In a fit of inspiration, here's a suggestion. This would be placed after the last sentence of the 2nd paragraph in the 3-paragraph description.
"This community involvement allows developers to focus on what's most important to them in the knowledge that the community will provide high quality packages for everything else."
Maybe there should be added that Arch Linux is only a distro from its devs for its devs and not from the community for the community.
I wish you wouldn't add such a paragraph. Instead the ignorance of some devs should be removed. That was the better way.
And Heiko, I'm not sure why you're making a big fuss over these packages. Its not even much harder to work from the AUR. Comparisons to Gentoo are ridiculous. I've got 75 packages from the AUR currently, and I don't spend hours a day compiling. The longest package to compile is probably samba4 at 20 minutes, followed by the kernel at 10 minutes (optimized), everything else tops out at 1-2 minutes. Go have a cup of coffee, and consider the difference to the hours and hours you'd have spent compiling world on Gentoo.
I already used Gentoo for 6 years. So I know what that compiling means. That's why I was looking for a binary distro in the style of Gentoo. That's why I came to Arch Linux 3 or 4 years ago.
And the more small packages need to be compiled from AUR the more time it takes compiling them. Many a little makes a mickle.
It's not at all about compile times IMHO but about the effort necessary to keeping source packages up-to-date. Doing this manually would be a major PITA for even a relatively small number of packages. AUR helpers like slurpy help a bit, but the required effort is still significantly higher than with binary packages. How much effort it requires depends IMHO almost entirely on the number of packages. JFYI I currently maintain 89 AUR packages and have probably about 180 AUR packages from other maintainers installed. There is does mean effort, believe me or not.
My main problems are this mass cleanup and the ignorant responses I get from some devs to such remarks or to some bug reports etc.
That's why I make such a fuss.
Heiko
On Tue 16 Nov 2010 13:06 +0100, Heiko Baums wrote:
Am Tue, 16 Nov 2010 21:54:23 +1000 schrieb Allan McRae <allan@archlinux.org>:
It is true as long as the developers say it is true. Just because I tell GM their car is an aeroplane does not mean they will start flying.
Wrong. It is true as soon as a distro becomes more and more popular. And this is the case for Arch Linux. As long as a distro is unknown and only used by a few people, mainly its developers, this distro may only be from the devs for the devs. But as soon as it is mentioned together with and equivalent to the other big distros and gets more popular this is not true anymore independent from what a single developer says.
Otherwise you should write a big note on the homepage and/or the download page that this distro is free but not meant to be used for the public. "Feel free to use it but don't expect anything. This distro is only meant to be used by us developers." Or something like this. Maybe a bit exaggerated.
You're right. A lot of open source software does have that no guarantee, no liability disclaimer. Maybe it should be made more obvious. Anyways, Arch Linux isn't bound by any unwritten or unspoken contracts saying that it must deliver a certain level of support after reaching a certain popularity. If you want that support, then go a head and contribute the resources. I really don't forsee the distro signing any support contracts though.
Am Tue, 16 Nov 2010 07:36:59 -0500 schrieb Loui Chang <louipc.ist@gmail.com>:
You're right. A lot of open source software does have that no guarantee, no liability disclaimer. Maybe it should be made more obvious.
Anyways, Arch Linux isn't bound by any unwritten or unspoken contracts saying that it must deliver a certain level of support after reaching a certain popularity. If you want that support, then go a head and contribute the resources. I really don't forsee the distro signing any support contracts though.
I'm not talking about contracts. But regardless of any contracts developers shouldn't ignore user's comments and opinions. I'm just talking about such an ignorant and sometimes arrogant behaviour of some devs which is seen sometimes in flyspray and on the mailing list. This makes using a distro and contributing to a distro no fun. And I haven't seen such an ignorance in other distros, yet. That's the point. Heiko
On Tuesday 16 November 2010 12:48:03 Heiko Baums wrote:
Man, you know that you have removed most of the packages from your "draft" from [extra]. Man, you know my list was made by 352 packages and I removed only ~180. Not every package in that list will be removed so calm down.
And I need to install three times as many packages from AUR as I needed three years ago just because they have been removed from the repos to AUR. This is the third or fourth time that you repeat this. Do you think that I care about it? I already told you: if you want to add those packages, apply as TU. You are welcome. If you cannot and if you have not the time to build an updated version of a package you should think about switch distro.
And the repos of Arch Linux are pretty small compared to other distros anyway. That's always mentioned as disadvantage in reviews of Arch Linux. This is a bit compensated by AUR, but if you regularly move important or popular packages to AUR then Arch Linux will become a second Gentoo and for some people who want or need a binary distro unusable. I don't care about comparisons with others distro. Let people use what they need, not the distro with more binaries.
-- Andrea Scarpino Arch Linux Developer
On Tue, Nov 16, 2010 at 1:23 PM, Andrea Scarpino <andrea@archlinux.org> wrote:
On Tuesday 16 November 2010 12:48:03 Heiko Baums wrote: if you want to add those packages, apply as TU. You are welcome. If you cannot and if you have not the time to build an updated version of a package you should think about switch distro.
Hello, I am a new user of arch since 3 months. I take this opportunity, between 2 shots, to propose to adopt package squashfs-tools. But i'm not a TU and i believe, it's necessary. So i need a TU mentor. Regards, -- Sébastien Luttringer www.seblu.net
As already mentioned, many of these packages haven't been updated upstream in a long time, and I doubt many of them will. For example: Name: polymer Upstream: http://static.int.pl/~mig21/dev/releases/polymer/ Description: "QT3 port of Plastik" Last news update: "14.05.2005" Really? 2005 and QT3? Moral of the story: just adopt them. Chances are you will never have to do any work with them (excluding the ones that *actually* do get upstream changes), and it makes the repo look/function in a whole lot cleaner manner. Regards, Brad
As already mentioned, many of these packages haven't been updated upstream in a long time, and I doubt many of them will.
For example: Name: polymer Upstream: http://static.int.pl/~mig21/dev/releases/polymer/<http://static.int.pl/%7Emig21/dev/releases/polymer/> Description: "QT3 port of Plastik" Last news update: "14.05.2005"
Really? 2005 and QT3?
Moral of the story: just adopt them. Chances are you will never have to do any work with them (excluding the ones that *actually* do get upstream changes), and it makes the repo look/function in a whole lot cleaner manner
I think it's kind of hard for me to see why I should maintain a package that's already been discarded by its developer. In my opinion such packages should be moved to [unsupported] where the one more two people who might want to use them can simply build them themselves. --Kaiting. -- Kiwis and Limes: http://kaitocracy.blogspot.com/
Am Tue, 16 Nov 2010 23:19:40 -0500 schrieb Kaiting Chen <kaitocracy@gmail.com>:
I think it's kind of hard for me to see why I should maintain a package that's already been discarded by its developer. In my opinion such packages should be moved to [unsupported] where the one more two people who might want to use them can simply build them themselves.
Why should those packages be removed from the repos as long as they are running? That doesn't make sense. And such packages doesn't make any work for the developers. They can just be staying in the repos without doing any harm like e.g. eboard. Regarding ding as an example doesn't make much work for the devs because it's updated by upstream every two years. And this package is really popular at least in Germany, because it's an English-German dictionary. And this tool is really old - but not outdated and unmainted. It's one of the first Linux applications and available in every repo of every distro. And the question is not cleaning up the repos in principle. The question is this mass cleanup and the removal of several popular and important packages even if they are orphaned. If there's an orphan quite popular then an unorphaned packages which is not popular or important could be moved to AUR and the orphaned and more popular package could be adopted by this dev. Just an example. squashfs-tools are necessary for building LiveCDs incl. the Arch Linux installation CD as far as I know. So I'm not sure if this package actually wouldn't belong to [core]. btrfs-progs also doesn't belong to AUR. This package belongs into [core] and should be supported by AIF. Even if it's still marked as experimental, many people in the web report that it's pretty stable and that it's only missing an fsck. And many people report that it's usable on systems which don't need to be absolutely reliable. Btw., instead of the stable package btrfs-progs there's a package btrfs-progs-unstable in [extra] which really makes sense as the repos are meant to be stable repos. eboard, a still working and good chess GUI, was moved from [extra] to AUR. It's not maintained by upstream anymore but it's still working, it's quite popular and doesn't make any work for the devs. Having this in [extra] means there's a compiled and working package which doesn't need to be maintained. Having this package in AUR means that every user who wants to install this package must compile this package by himself. So what sense does this cleanup make? It makes completely no sense! epdfviewer is a very popular because lightweight PDF viewer for GTK. Galculator is the best calculator for GTK I know and also quite popular, at lest recommended quite often e.g. in the Xfce wiki. What's such a package doing in AUR? And, please, don't tell me anything about missing interest of the devs. As if every dev is using every package which he maintains himself or every dev only maintains only packages he is using himself. This is what I name and shame. This mass cleanup was just done inconsiderately. I really respect the voluntary work of the devs and TUs. And I really honor their work in their spare time. And I don't expect too much. But if a repo shall be cleaned up this must be done a lot more considered. Heiko
On 17/11/10 22:45, Heiko Baums wrote:
Am Tue, 16 Nov 2010 23:19:40 -0500 schrieb Kaiting Chen<kaitocracy@gmail.com>:
I think it's kind of hard for me to see why I should maintain a package that's already been discarded by its developer. In my opinion such packages should be moved to [unsupported] where the one more two people who might want to use them can simply build them themselves.
Why should those packages be removed from the repos as long as they are running? That doesn't make sense. And such packages doesn't make any work for the developers. They can just be staying in the repos without doing any harm like e.g. eboard.
Because there is no-one in charge of any bug reports, monitoring security issues, rebuilding the package for soname bumps... Packages without a maintainer do cause all other devs needless work.
Regarding ding as an example doesn't make much work for the devs because it's updated by upstream every two years. And this package is really popular at least in Germany, because it's an English-German dictionary. And this tool is really old - but not outdated and unmainted. It's one of the first Linux applications and available in every repo of every distro.
Every distro... bold statement! http://chakra-project.org/packages/index.php?subdir=&sortby=name&order=ascending&act=search&searchpattern=ding Not there....
And the question is not cleaning up the repos in principle. The question is this mass cleanup and the removal of several popular and important packages even if they are orphaned.
If there's an orphan quite popular then an unorphaned packages which is not popular or important could be moved to AUR and the orphaned and more popular package could be adopted by this dev. Just an example.
Why would a dev drop a package they use and actively maintain for another one they do not use? That seems no fun, and given fun is what motivates volunteers...
squashfs-tools are necessary for building LiveCDs incl. the Arch Linux installation CD as far as I know. So I'm not sure if this package actually wouldn't belong to [core].
It is not needed to boot your system, so it definitely does not belong in [core]. None of the release engineering team have mentioned that it is needed either...
btrfs-progs also doesn't belong to AUR. This package belongs into [core] and should be supported by AIF. Even if it's still marked as experimental, many people in the web report that it's pretty stable and that it's only missing an fsck. And many people report that it's usable on systems which don't need to be absolutely reliable.
Btw., instead of the stable package btrfs-progs there's a package btrfs-progs-unstable in [extra] which really makes sense as the repos are meant to be stable repos.
Agreed. Upstream labels it unstable software so both should be dropped if we are consistent. This is just getting rid of more packages! :P
eboard, a still working and good chess GUI, was moved from [extra] to AUR. It's not maintained by upstream anymore but it's still working, it's quite popular and doesn't make any work for the devs. Having this in [extra] means there's a compiled and working package which doesn't need to be maintained. Having this package in AUR means that every user who wants to install this package must compile this package by himself. So what sense does this cleanup make? It makes completely no sense!
It still needs maintained... see above. If it was really no issue to build it once, then why complain if it is in the AUR. Or use xboard, pychess, etc that are used by developers here and maintained in our repos.
epdfviewer is a very popular because lightweight PDF viewer for GTK. Galculator is the best calculator for GTK I know and also quite popular, at lest recommended quite often e.g. in the Xfce wiki. What's such a package doing in AUR?
The wiki also recommends using an AUR helper and they are very popular, but we do not put those in the repos.
And, please, don't tell me anything about missing interest of the devs. As if every dev is using every package which he maintains himself or every dev only maintains only packages he is using himself.
I use every package I maintain and follow upstream mailing lists for most of them. Maintaining a package well is more than just dumping a binary in the repo. Allan
2010/11/17 Allan McRae <allan@archlinux.org>:
On 17/11/10 22:45, Heiko Baums wrote:
Am Tue, 16 Nov 2010 23:19:40 -0500 schrieb Kaiting Chen<kaitocracy@gmail.com>:
I think it's kind of hard for me to see why I should maintain a package that's already been discarded by its developer. In my opinion such packages should be moved to [unsupported] where the one more two people who might want to use them can simply build them themselves.
Why should those packages be removed from the repos as long as they are running? That doesn't make sense. And such packages doesn't make any work for the developers. They can just be staying in the repos without doing any harm like e.g. eboard.
Because there is no-one in charge of any bug reports, monitoring security issues, rebuilding the package for soname bumps... Packages without a maintainer do cause all other devs needless work.
Why not move them to a "graveyard" repo, that would be called [unmaintained]. It would contain binary packages that belonged formerly to [community] but are explicitly not maintained anymore. That would allow people who use them to still have binary packages, until it doesn't work anymore (then someone files an out-of-date notice and the package has to be deleted). -- Rémy.
On Wednesday 17 November 2010 15:51:22 Rémy Oudompheng wrote:
Why not move them to a "graveyard" repo, that would be called [unmaintained]. It would contain binary packages that belonged formerly to [community] but are explicitly not maintained anymore. That would allow people who use them to still have binary packages, until it doesn't work anymore (then someone files an out-of-date notice and the package has to be deleted). Looking for an old binary? http://schlunix.org/?page_id=11 http://arm.konnichi.com/search/
Seriously, stop this thread goes nowhere. The only way to keep these packages in [community] is to adopt them. -- Andrea Scarpino Arch Linux Developer
Excerpts from Heiko Baums's message of 2010-11-16 12:48:03 +0100:
Am Tue, 16 Nov 2010 12:12:23 +0100 schrieb Andrea Scarpino <andrea@archlinux.org>:
Man you know that is a draft. I don't remember who wants to adopt squashfs- tools now, but surely a TU want because it has been moved to [community] recently.
Man, you know that you have removed most of the packages from your "draft" from [extra].
And I need to install three times as many packages from AUR as I needed three years ago just because they have been removed from the repos to AUR.
And the repos of Arch Linux are pretty small compared to other distros anyway. That's always mentioned as disadvantage in reviews of Arch Linux. This is a bit compensated by AUR, but if you regularly move important or popular packages to AUR then Arch Linux will become a second Gentoo and for some people who want or need a binary distro unusable.
The problem with Arch becoming a second gentoo is that it would be a far worse gentoo. AUR isn't exactly convenient. Sure there are helper programs, but each one I tried buggy or lacking. I doubt maintaining source packages in gentoo is as much a PITA as it is in Arch. It's less a PITA in Arch than in binary distros, but still only really usable as long as Arch is mainly a binary distro. My point in short: Arch is great as long most packages you need are binaries and only some are from source. If Arch requires you to build lots of packages from source it's the worst of both worlds.
And no, Arch Linux is not a distro from developers for developers anymore. This may have been in the beginning or Arch Linux but is not true anymore.
Heiko
On Tue, 2010-11-16 at 16:08 +0100, Philipp Überbacher wrote:
My point in short: Arch is great as long most packages you need are binaries and only some are from source. If Arch requires you to build lots of packages from source it's the worst of both worlds.
Have you seen the list of software getting moved? I've only seen one person describe them collectively as 'many important' packages. Almost none of which I've heard of before, of course.... This thread started with the assertion that 'many important' packages are getting moved to the AUR. I believe this assertion to be false, as, obviously, do the devs. Historically from reading [arch-dev-public] the devs have been careful to continue maintaining packages none of them use if its seen as crucial to a large majority of users. None of the packages being moved fit these criteria. At all.
Excerpts from Ng Oon-Ee's message of 2010-11-16 17:13:47 +0100:
On Tue, 2010-11-16 at 16:08 +0100, Philipp Überbacher wrote:
My point in short: Arch is great as long most packages you need are binaries and only some are from source. If Arch requires you to build lots of packages from source it's the worst of both worlds.
Have you seen the list of software getting moved? I've only seen one person describe them collectively as 'many important' packages. Almost none of which I've heard of before, of course....
This thread started with the assertion that 'many important' packages are getting moved to the AUR. I believe this assertion to be false, as, obviously, do the devs. Historically from reading [arch-dev-public] the devs have been careful to continue maintaining packages none of them use if its seen as crucial to a large majority of users. None of the packages being moved fit these criteria. At all.
I see your point. I looked through the [extra] -> [] list yesterday or so and was a bit shocked at first until I saw that most of the packages I considered important would be maintained in community. However, what I tried to point out is what would happen if a binary -> source trend develops. One other thing: The lists are based on orphans. My impression was that it's common practice among developers to adopt -> update -> orphan. Based on this I wonder whether it's sensible to create lists of removal candidates based on orphans.
On Tue, 2010-11-16 at 18:28 +0100, Philipp Überbacher wrote:
Excerpts from Ng Oon-Ee's message of 2010-11-16 17:13:47 +0100:
This thread started with the assertion that 'many important' packages are getting moved to the AUR. I believe this assertion to be false, as, obviously, do the devs. Historically from reading [arch-dev-public] the devs have been careful to continue maintaining packages none of them use if its seen as crucial to a large majority of users. None of the packages being moved fit these criteria. At all.
I see your point. I looked through the [extra] -> [] list yesterday or so and was a bit shocked at first until I saw that most of the packages I considered important would be maintained in community.
However, what I tried to point out is what would happen if a binary -> source trend develops.
No danger of that happening here. The devs use Arch, why would they move out stuff they consider important? And between them, what they consider important is probably almost everything in a 'base' install for the big distros (Ubuntu, Fedora etc.), because importance also depends on user-base. Higher user-base means its more likely a dev uses it =).
One other thing: The lists are based on orphans. My impression was that it's common practice among developers to adopt -> update -> orphan. Based on this I wonder whether it's sensible to create lists of removal candidates based on orphans.
I think that's only common practice for packages where noone really wants them, the initial dev has left/become inactive, so its in a state of limbo, but when devs have the time they look at the 'out-of-date' flag and help out.
Before anyone gets any more worked up I wanted to point out that there are only 27 Trusted Users and 32 Developers. In the official repository there are around 4848 packages altogether. That averages out to 82.169 packages per person which is kind of ridiculous considering the Developers have to develop and the Trusted Users have other responsibilities as well. So I think this binary -> source phenomenon is just the best that we can do given how shorthanded we all are. I think the philosophy is that anyone can adopt in [unsupported], and if no Developer or Trusted User will adopt a binary package then we should at least give the concerned user a chance to adopt it. Honestly if it were up to me I would remove half of the packages from the official repositories and stick them in the AUR because past 40 your packages start to develop major Quantity over Quality issues and I don't think that's what we're going for. I think this would be a much more constructive conversation if we all stopped complaining about the situation and started talking about how to improve the AUR. --Kaiting. -- Kiwis and Limes: http://kaitocracy.blogspot.com/
On Tue, Nov 16, 2010 at 9:04 PM, Kaiting Chen <kaitocracy@gmail.com> wrote:
Before anyone gets any more worked up I wanted to point out that there are only 27 Trusted Users and 32 Developers. In the official repository there are around 4848 packages altogether. That averages out to 82.169 packages per person which is kind of ridiculous considering the Developers have to develop and the Trusted Users have other responsibilities as well.
So I think this binary -> source phenomenon is just the best that we can do given how shorthanded we all are. I think the philosophy is that anyone can adopt in [unsupported], and if no Developer or Trusted User will adopt a binary package then we should at least give the concerned user a chance to adopt it.
Honestly if it were up to me I would remove half of the packages from the official repositories and stick them in the AUR because past 40 your packages start to develop major Quantity over Quality issues and I don't think that's what we're going for.
I think this would be a much more constructive conversation if we all stopped complaining about the situation and started talking about how to improve the AUR. --Kaiting.
-- Kiwis and Limes: http://kaitocracy.blogspot.com/
That's not a very good argument. Sergej Pupykin: 1480 packages Jan de Groot: 1094 packages Andrea Scarpino: 809 packages They all do an excellent job with maintaining a massive amount of packages at one time. Therefore, it obviously can be done without the "quantity over quality" issue that you speak of. Regards, Brad
On Wed, Nov 17, 2010 at 1:21 AM, Brad Fanella <bradfanella@archlinux.us> wrote:
That's not a very good argument.
Sergej Pupykin: 1480 packages Jan de Groot: 1094 packages Andrea Scarpino: 809 packages
They all do an excellent job with maintaining a massive amount of packages at one time. Therefore, it obviously can be done without the "quantity over quality" issue that you speak of.
O.O Man, I feel so depressed now, serious... I'm so unproductive and unorganized... Please tell me the that this 3 people are something like Borgs made of hundreds of developers minds linked together... =] (Very best) Regards, Kazuo -- “The journey is more important than the destination—that’s part of life, if you only live for getting to the end, you’re almost always disappointed.” Donald E. Knuth
On Tue 16 Nov 2010 21:21 -0600, Brad Fanella wrote:
On Tue, Nov 16, 2010 at 9:04 PM, Kaiting Chen <kaitocracy@gmail.com> wrote:
Honestly if it were up to me I would remove half of the packages from the official repositories and stick them in the AUR because past 40 your packages start to develop major Quantity over Quality issues and I don't think that's what we're going for.
I think this would be a much more constructive conversation if we all stopped complaining about the situation and started talking about how to improve the AUR. --Kaiting.
That's not a very good argument.
Sergej Pupykin: 1480 packages Jan de Groot: 1094 packages Andrea Scarpino: 809 packages
They all do an excellent job with maintaining a massive amount of packages at one time. Therefore, it obviously can be done without the "quantity over quality" issue that you speak of.
I think he was talking about mere mortals when he wrote that. Kaiting has a very good point though that much could be done to improve the AUR making it easier to deal with source packages and maybe other source based repos in Arch.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 17/11/10 05:21, Brad Fanella wrote:
On Tue, Nov 16, 2010 at 9:04 PM, Kaiting Chen <kaitocracy@gmail.com> wrote:
Before anyone gets any more worked up I wanted to point out that there are only 27 Trusted Users and 32 Developers. In the official repository there are around 4848 packages altogether. That averages out to 82.169 packages per person which is kind of ridiculous considering the Developers have to develop and the Trusted Users have other responsibilities as well.
So I think this binary -> source phenomenon is just the best that we can do given how shorthanded we all are. I think the philosophy is that anyone can adopt in [unsupported], and if no Developer or Trusted User will adopt a binary package then we should at least give the concerned user a chance to adopt it.
Honestly if it were up to me I would remove half of the packages from the official repositories and stick them in the AUR because past 40 your packages start to develop major Quantity over Quality issues and I don't think that's what we're going for.
I think this would be a much more constructive conversation if we all stopped complaining about the situation and started talking about how to improve the AUR. --Kaiting.
-- Kiwis and Limes: http://kaitocracy.blogspot.com/
That's not a very good argument.
Sergej Pupykin: 1480 packages Jan de Groot: 1094 packages Andrea Scarpino: 809 packages
They all do an excellent job with maintaining a massive amount of packages at one time. Therefore, it obviously can be done without the "quantity over quality" issue that you speak of.
I strongly agree with Kaiting. In my opinion, more than 50-60 packages is pushing it. I doubt the numbers you posted are accurate. For example, (and if I'm not mistaken,) Andrea maintains lots of KDE packages which are split into many sub-packages. The number of actual PKGBUILDs is significantly smaller. This fact makes maintaining all these (sub-)packages much easier. For more precise measurements, you could grep through all PKGBUILDs in /var/abs: $ grep -r '# Maintainer' /var/abs | grep -c Scarpino 152 $ grep -r '# Maintainer' /var/abs | grep -c Groot 456 $ grep -r '# Maintainer' /var/abs | grep -c Pupykin 708 Moreover, for a conservative maintenance time estimation of 10 minutes/package/month, you will need to spend 10 hours each month, updating and solving bugs, for every 60 packages you own. This doesn't scale very well; if you had 300 packages, you'd need to set aside 50 hours each month. Let's not forget we're talking about orphan packages here. This means that nobody is interested in maintaining them. Why exactly should they stay in [community], when someone who cares about them can adopt and maintain them properly in the AUR? - -- Evangelos Foutras -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) iQEcBAEBAgAGBQJM41XWAAoJEFHosUipmZw0basH/0tYeoV/IWSh3TxcaCx6pW8q /hBA9A1nHdsyiCRAAHB0pRGe80ELqukmTQyxysKXcmgHxWmjkIM0wvpZXgVAt3ip 6CQf3S6WXo0Ly4glVnQGmj/JhJsLOq27hGJu8YX3BG9cp6cmyJHUefo0TvgnoRrq FauM7smwfhxeOmLOJuQWlaPIgIl58lHx8nLSdfseGY0XMVVETNJHIpo1UXFNIsql NJG4ZTjKJvU8xknJnbshdwY3/E55GUaaksUyEp55cnRj0lWHfqpMSgZ4vjApyqSW pgE8rEK7S8XxkvJiPyoKp82l+V4qW9jtoCMuIiWG2ibGTnwupzEIOEO2zjZHSE8= =ZzEW -----END PGP SIGNATURE-----
Hi, I think there is no need to move orphaned package to AUR until it works and does not have critical/security bugs.
On 16 November 2010 15:21, Sergej Pupykin <ml@sergej.pp.ru> wrote:
Hi,
I think there is no need to move orphaned package to AUR until it works and does not have critical/security bugs.
+1. Why remove working packages?
On Tue, 2010-11-16 at 16:27 +0100, Lukáš Jirkovský wrote:
On 16 November 2010 15:21, Sergej Pupykin <ml@sergej.pp.ru> wrote:
Hi,
I think there is no need to move orphaned package to AUR until it works and does not have critical/security bugs.
+1. Why remove working packages?
If unmaintained, then why not let someone who actually uses the package maintain it? Its not like the packages are gone forever. The exact same PKGBUILDs are now on the AUR, so you can still use the packages. Or update them if you're interested. The only 'loss' is to those who don't have a package installed, they'd now have to install something from the AUR. Big deal.
At Wed, 17 Nov 2010 00:15:44 +0800, Ng Oon-Ee <ngoonee@gmail.com> wrote:
On Tue, 2010-11-16 at 16:27 +0100, Lukáš Jirkovský wrote:
On 16 November 2010 15:21, Sergej Pupykin <ml@sergej.pp.ru> wrote:
Hi,
I think there is no need to move orphaned package to AUR until it works and does not have critical/security bugs.
+1. Why remove working packages?
If unmaintained, then why not let someone who actually uses the package maintain it?
I prefer working binary out-of-date package than up-to-date which should be compiled from AUR.
On Tuesday 16 November 2010 17:51:44 Sergej Pupykin wrote:
I prefer working binary out-of-date package than up-to-date which should be compiled from AUR. An up-to-date package includes upstream fixes, so I prefer an up-to-date PKGBUILD from AUR.
-- Andrea Scarpino Arch Linux Developer
At Tue, 16 Nov 2010 18:02:35 +0100, Andrea Scarpino wrote:
On Tuesday 16 November 2010 17:51:44 Sergej Pupykin wrote:
I prefer working binary out-of-date package than up-to-date which should be compiled from AUR. An up-to-date package includes upstream fixes, so I prefer an up-to-date PKGBUILD from AUR.
I mean only well working packages, so no critical fixes needed.
On Wed, 2010-11-17 at 00:15 +0800, Ng Oon-Ee wrote:
On Tue, 2010-11-16 at 16:27 +0100, Lukáš Jirkovský wrote:
On 16 November 2010 15:21, Sergej Pupykin <ml@sergej.pp.ru> wrote:
Hi,
I think there is no need to move orphaned package to AUR until it works and does not have critical/security bugs.
+1. Why remove working packages?
If unmaintained, then why not let someone who actually uses the package maintain it?
I actually use python-pyparallel, which is why I stuck my name beside it. Funny that python-pyserial (closely related) is still maintained by the original packager... Gordon MM0YEQ
On 16 November 2010 19:03, Heiko Baums <lists@baums-on-web.de> wrote:
Am Tue, 16 Nov 2010 11:34:11 +0100 schrieb Andrea Scarpino <andrea@archlinux.org>:
Hi TUs, we are here again. After the success with [extra] (DEVs adopted ~80 packages, TUs ~60), I want to reduce the number of orphans packages in [community]. Actually, they are 84 (82, I just adopted two...).
The list is here[1]. Simply cross out the package which you want to maintain in [community]. Adoption is not required, but would be nice. Packages will be moved to AUR this Saturday 20th.
Thanks
[1] https://wiki.archlinux.org/index.php/DeveloperWiki:Community_Repo_Cleanup
squashfs-tools
What have you told me about that when you have moved this from [extra] to [community]?
Now it shall be moved to AUR? And why was it moved to [community] if it is an orphan there? And isn't this needed to building LiveCDs including the Arch Linux installation CDs?
You really should think about that.
Man, you are right. In fact, I was just studying archiso and squashfs-tools is one of the dependencies. If no-one is going to maintain it, I will :) See, as long as it's important and there is someone on the team who actually needs it, it'll be fine. I was a Gentoo user just like you, and came to Arch Linux for exactly the same reasons. I do not think it is straying away just because some orphaned packages are being "dropped" to AUR. So, come on board and save them! Even if something's dropped, it's not like it can't be adopted back if and when deemed important. It is not the end! You see, the way Arch started and the way it is now is not going to be any different just because it has "grown" popular. I think you should know this very well, being an Archer for some time. This is not the first time we have had disagreements between users (a developer/TU is a user). For some historical fun, google the term 'archmilkers'.
On Tuesday 16 November 2010 11:34:11 Andrea Scarpino wrote:
Hi TUs, we are here again. After the success with [extra] (DEVs adopted ~80 packages, TUs ~60), I want to reduce the number of orphans packages in [community]. Actually, they are 84 (82, I just adopted two...).
The list is here[1]. Simply cross out the package which you want to maintain in [community]. Adoption is not required, but would be nice. Packages will be moved to AUR this Saturday 20th.
Thanks
[1] https://wiki.archlinux.org/index.php/DeveloperWiki:Community_Repo_Cleanup No a reply or an edit in 24h. I am starting to move these 41 packages to AUR.
-- Andrea Scarpino Arch Linux Developer
participants (19)
-
Allan McRae
-
Andrea Scarpino
-
Andreas Radke
-
Brad Fanella
-
Evangelos Foutras
-
Gordon JC Pearce
-
Heiko Baums
-
Ionuț Bîru
-
Kaiting Chen
-
Kazuo Teramoto
-
Loui Chang
-
Lukáš Jirkovský
-
Ng Oon-Ee
-
Peter Lewis
-
Philipp Überbacher
-
Ray Rashif
-
Rémy Oudompheng
-
Seblu
-
Sergej Pupykin