[arch-general] makepkg as root
I just noticed that `makepkg` no longer accepts '--asroot', does that mean there now is no way to convince `makepkg` to build despite being run as root? /M P.S. I am fully aware of the problems with building as root, but as I'm only building packages I'm myself creating, and doing it in a docker image, I feel I can live with the dangers... especially when it saves me some work in setting up docker images. -- Magnus Therning OpenPGP: 0xAB4DFBA4 email: magnus@therning.org jabber: magnus@therning.org twitter: magthe http://therning.org/magnus What gets measured, gets done. -- Tom Peters
On 12/30/14, Magnus Therning <magnus@therning.org> wrote:
I just noticed that `makepkg` no longer accepts '--asroot', does that mean there now is no way to convince `makepkg` to build despite being run as root?
/M
P.S. I am fully aware of the problems with building as root, but as I'm only building packages I'm myself creating, and doing it in a docker image, I feel I can live with the dangers... especially when it saves me some work in setting up docker images.
-- Magnus Therning OpenPGP: 0xAB4DFBA4 email: magnus@therning.org jabber: magnus@therning.org twitter: magthe http://therning.org/magnus
What gets measured, gets done. -- Tom Peters
Building as root was disabled due to other changes.[1] It seems your use-case was either not anticipated or ignored. But you aren't the only one who got inconvenienced. Someone in the comments is building packages in a chroot on android which apparently has issues with fakeroot. Allan McRae suggested recompiling fakeroot with different options; I imagine if there was some way to trick makepkg into running as root, he'd've suggested that instead. [1] http://allanmcrae.com/2014/12/pacman-4-2-released/ -- Eli Schwartz
I am/was an '--asroot' user too. Does a feaure request makes sense? Happy New Year!
Don't think you'll get it added back officially. See the discussion in Allan's blog: http://allanmcrae.com/2014/12/pacman-4-2-released/ On Fri, Jan 2, 2015 at 8:09 AM, Ralf Mardorf <ralf.mardorf@rocketmail.com> wrote:
I am/was an '--asroot' user too. Does a feaure request makes sense?
Happy New Year!
'--asroot' was removed with deliberation and forethought. If anyone can think of a compelling reason the developers have not thought of, perhaps you will have luck convincing them. Random feature requests saying you want it probably don't qualify... at least explain why you are not able to simply run 'makepkg' in another user account. I do have to wonder, though, why... one is running as root by default??? Why else would '--asroot' be needed? -- Eli Schwartz
On Fri, 2 Jan 2015 13:35:52 -0500, Eli Schwartz wrote:
'--asroot' was removed with deliberation and forethought.
If anyone can think of a compelling reason the developers have not thought of, perhaps you will have luck convincing them. Random feature requests saying you want it probably don't qualify... at least explain why you are not able to simply run 'makepkg' in another user account.
I do have to wonder, though, why... one is running as root by default??? Why else would '--asroot' be needed?
In the past I sometimes had no reason _not_ to run makepkg as root, when I wanted to compile from ABS, located in /var/abs/. I'm aware about the drawback, that actually isn't relevant for my kind of computer usage + several available backups. However, indeed, for my needs non-root 100% does do the job too. No, I won't copy from ABS to some user dir, I simply will chown or chmod in a way the big teachers won't us to act self-responsible (I simply ignored the kindergarten flame-war like posts). JFTR my machine isn't a server or a terrorist top secret whatsoever machine. It's a digital audio workstation and there are several, individual, complete backups available for each month. Resume: I can live without '--asroot', but I'm against dropping it, just for backwards compatibility. Imagine that somebody wrote a script or whatever, that needs '--asroot'. IMOH it isn't wise to ignore backwards compatibility. And btw. I dislike the claims mentioned by a link ... "Linux has never been about ‘choice’ or ‘freedom’ and those myths should just die out." ... underpinned with a link to redhat :(. IMHO Johannes Löthberg is completely wrong, likely a maintainer for the wrong distro.
On Fri, Jan 02, 2015 at 08:12:12PM +0100, Ralf Mardorf wrote:
In the past I sometimes had no reason _not_ to run makepkg as root, when I wanted to compile from ABS, located in /var/abs/. I'm aware about the drawback, that actually isn't relevant for my kind of computer usage + several available backups. However, indeed, for my needs non-root 100% does do the job too. No, I won't copy from ABS to some user dir, I simply will chown or chmod in a way the big teachers won't us to act self-responsible (I simply ignored the kindergarten flame-war like posts). JFTR my machine isn't a server or a terrorist top secret whatsoever machine. It's a digital audio workstation and there are several, individual, complete backups available for each month.
ABSROOT=~/abs abs You can set ABSROOT in your bashrc or zshrc.
On Fri, 2 Jan 2015 11:28:05 -0800, Ian D. Scott wrote:
On Fri, Jan 02, 2015 at 08:12:12PM +0100, Ralf Mardorf wrote:
In the past I sometimes had no reason _not_ to run makepkg as root, when I wanted to compile from ABS, located in /var/abs/. I'm aware about the drawback, that actually isn't relevant for my kind of computer usage + several available backups. However, indeed, for my needs non-root 100% does do the job too. No, I won't copy from ABS to some user dir, I simply will chown or chmod in a way the big teachers won't us to act self-responsible (I simply ignored the kindergarten flame-war like posts). JFTR my machine isn't a server or a terrorist top secret whatsoever machine. It's a digital audio workstation and there are several, individual, complete backups available for each month.
ABSROOT=~/abs abs
You can set ABSROOT in your bashrc or zshrc.
Oops :D. My apologies :). Thank you, fair play :D. I'll test it soon. Regards, Ralf
On Fri, Jan 2, 2015 at 2:12 PM, Ralf Mardorf <ralf.mardorf@rocketmail.com> wrote:
In the past I sometimes had no reason _not_ to run makepkg as root, when I wanted to compile from ABS, located in /var/abs/. I'm aware about the drawback, that actually isn't relevant for my kind of computer usage + several available backups. However, indeed, for my needs non-root 100% does do the job too. No, I won't copy from ABS to some user dir, I simply will chown or chmod in a way the big teachers won't us to act self-responsible (I simply ignored the kindergarten flame-war like posts). JFTR my machine isn't a server or a terrorist top secret whatsoever machine. It's a digital audio workstation and there are several, individual, complete backups available for each month.
If you are using yaourt, there is a sync option -b, --build Build from sources, ABS for official packages, or AUR if packages is not found. Specify this option twice to build all dependencies. So: 'yaourt -Sb packagename' should build that package even if it is from the repos.You don't even need to keep the abs tree. Just build stuff the same way you'd build AUR packages.
Resume: I can live without '--asroot', but I'm against dropping it, just for backwards compatibility. Imagine that somebody wrote a script or whatever, that needs '--asroot'.
IMOH it isn't wise to ignore backwards compatibility. And btw. I dislike the claims mentioned by a link ... "Linux has never been about 'choice' or 'freedom' and those myths should just die out." ... underpinned with a link to redhat :(. IMHO Johannes Löthberg is completely wrong, likely a maintainer for the wrong distro.
Well, I couldn't care less about their nonsense, but I still don't see why that means we need backwards compatibility. Backwards compatibility is a good thing, but never a reason to hold back the software from what it could become. I can think of worse things than people needing to update their automation scripts. Especially if it is something Arch doesn't want them doing anyway. ;)
On Fri, 2 Jan 2015 20:53:30 +0100, Martti Kühne wrote: [snip] Calm down! At least I decided to use Arch and I don't quote Redhead opinions ;). Assumed somebody likes Redhead opinions and does contribute to Arch, I'm not against it. Regarding this thread at least one example was given, why --asroot is useful. 1. There might be better reasons to drop it ... 2. ... anyway, backwards compatibility isn't unimportant. You're likely much to young to understand, why backwards compatibility could be important, assumed in "mar77i", the "77" is for born in 1977 ;). On Fri, 2 Jan 2015 13:52:30 -0600, Eli Schwartz wrote:
I can think of worse things than people needing to update their automation scripts. Especially if it is something Arch doesn't want them doing anyway. ;)
In this case it's not an issue for me, but I guess there are valid reasons to care about backwards compatibility. Regards, Ralf
On Fri, 2 Jan 2015 13:52:30 -0600, Eli Schwartz wrote:
If you are using yaourt
I do, but regarding the posted link Yaourt seems to be the biggest evil ever. Ok, the opinion is based on at least one claim that is untrue, since there's no need to run yaourt with root privileges. A fair discussion, without wrongs seems to be impossible, so I guess we simply should stop any discussions about this issue. Pardon that I chimed in. I don't need to run makepkg as root for my needs, but for very good reasons my concern is pro backwards compatibility, even while it's completely unimportant for me, regarding to makepkg. Again, I don't care about the makepkg options, I care about the principle of backwards compatibility in general.
Again, I don't care about the makepkg options, I care about the principle of backwards compatibility in general.
idem, I was going to post about it.
On Fri, 02 Jan 2015 22:08:37 +0100, Yamakaky wrote:
Again, I don't care about the makepkg options, I care about the principle of backwards compatibility in general.
idem, I was going to post about it.
"Ralf Mardorf on 2015/01/03 at 3:44 PM said: Hi, don’t get me wrong, most of my scripts start with something similar to if [ “$(id -u)” = “0” ]; then printf “\nDon’t run this script as root :).\n\n”; exit 1; fi However, to claim that software using an option of makepkg is super-bad, because it does use this option and to quote an opinion from folks using another distro, especially if it’s unrelated to backwards compatibility, doesn’t make a wrong decision a good one. If our software is old, we notice some mistakes and it’s good not to continue a mistake. Anyway, revised software should be backwards compatible as much as possible. To drop an option, or even just renaming an option is a bad idea. I guess dropping ‘–asroot’ isn’t a big issue. It’s ok, but consider to keep other software really backwards compatible. It shouldn’t become a fashion to ignore backwards compatibility. Happy New Year! Ralf" - http://allanmcrae.com/2014/12/pacman-4-2-released/#comment-1268 I really wonder why explained opinions need to be attacked with insults. Regarding to quotes about Linux in general, Linus Torvalds might be a better source than Redhat biased mailing lists.
On Fri, 2 Jan 2015 20:12:12 +0100 Ralf Mardorf <ralf.mardorf@rocketmail.com> wrote:
IMOH it isn't wise to ignore backwards compatibility. And btw. I dislike the claims mentioned by a link ... "Linux has never been about ‘choice’ or ‘freedom’ and those myths should just die out." ... underpinned with a link to redhat :(. IMHO Johannes Löthberg is completely wrong, likely a maintainer for the wrong distro.
I'm going to disagree with you on both counts. While maintaining backward compatibility is a good thing, there is often a cost, and very often that cost is too high. In this case, maintaining cruft and additional code complexity in order to maintain a feature that should never be used anyway simply isn't worth it to the authors/maintainers of the software in question. Without going completely into the battle that can be started by the link you mentioned, what about the "choice and freedom" of the authors? Don't they have the "choice and freedom" to write their software as they see fit? This would certainly seem to trump the "choice and freedom" of those that contribute nothing. Doug
I'm going to disagree with you on both counts. While maintaining backward compatibility is a good thing, there is often a cost, and very often that cost is too high. In this case, maintaining cruft and additional code complexity in order to maintain a feature that should never be used anyway simply isn't worth it to the authors/maintainers of the software in question.
I precise my point : it's a backward incompatible release, so it's version number should be 5.0, and not 4.2. See [semver] (http://semver.org/).
On January 3, 2015 10:59:41 AM EST, Doug Newgard <scimmia@archlinux.info> wrote:
Without going completely into the battle that can be started by the link you mentioned, what about the "choice and freedom" of the authors? Don't they have the "choice and freedom" to write their software as they see fit? This would certainly seem to trump the "choice and freedom" of those that contribute nothing.
Doug
Removing asroot reduces flexibility. For instance, when making packages on a live install usb. You'd have to create a new account and su into it. It seems like a projection of morals to me. Just because it's not good practice to use root permissions when it's not necessary, doesn't mean it won't ever be useful. The argument for choice and freedom of the creators can also be successfully used to justify DRM. -- vixsomnis
On Sat, Jan 03, 2015 at 03:45:35PM -0500, Christian Demsar wrote:
Removing asroot reduces flexibility. For instance, when making packages on a live install usb. You'd have to create a new account and su into it.
Why? Just do what Arch LiveCD has been doing for ages (i.e. having the 'arch' user). And regardless of makepkg, using an only-root system (even read-only) is asking for trouble.
It seems like a projection of morals to me. Just because it's not good practice to use root permissions when it's not necessary, doesn't mean it won't ever be useful.
makepkg uses fakeroot, there is simply no need to be running as a superuser.
The argument for choice and freedom of the creators can also be successfully used to justify DRM.
Companies are free to produce whatever crap they want. Unless you are a FSF (or RMS) fanatic, there is nothing wrong with DRM really. Cheers, -- Leonid Isaev GPG fingerprints: DA92 034D B4A8 EC51 7EA6 20DF 9291 EE8A 043C B8C4 C0DF 20D0 C075 C3F1 E1BE 775A A7AE F6CB 164B 5A6D
On Sat, Jan 3, 2015 at 3:05 PM, Leonid Isaev <lisaev@umail.iu.edu> wrote:
Companies are free to produce whatever crap they want. Unless you are a FSF (or RMS) fanatic, there is nothing wrong with DRM really.
Cheers,
There is absolutely nothing right about denying people the right to use their own property however they like. I guess it is all right to want to make all new PCs sold with Windows 8 locked to only Windows 8? What about ebooks? I am a heavy reader, with lots of ebooks, but publishers don't want me to be able to read my books on linux, or on an ereader from a different vendor than the ebook came from. OK, yeah, I suppose I don't *have* to buy any of these things... but that doesn't seem like a reasonable option. If I change from a Kindle to a Kobo or Nook or Sony, I should have to repurchase all my ebooks? If the store I bought from goes out of business, I lose all future access to my ebooks? And the alternative is, I guess, don't read ebooks, just lug around a roomful of books everywhere I go? And while I am at it, I can do all my computation with the work of the late great Charles Babbage...
On Sat, 3 Jan 2015 09:59:41 -0600, Doug Newgard wrote:
Without going completely into the battle that can be started by the link you mentioned, what about the "choice and freedom" of the authors? Don't they have the "choice and freedom" to write their software as they see fit?
Linux and BSD user space is software based on other software. Yaourt and some other software does need pacman. Pacman needs bash, curl and other packages and those packages depend on other packages too. This policy does only work, when there are agreements about backwards compatibility. makepkg provided the --asroot option and within a major release it dropped backwards compatibility. Fortunately it is a minor issue that --asroot is dropped, but the real issue is that it already might be a fashion not to care about other software, not to care about work flows, IOW to be careless, ignorant. However, I like that Linus Torvalds made some clear statements to the careless, ignorant systemd and dbus crowed, while he's not against systemd and dbus, just against the new attitudes. Many of the older humans started using Linux, because other OS suffered from mistakes, that were avoided by Linux. It seems to be that a new generation now will introduce this mistakes to Linux (not the kernel, but the user space). Regards, Ralf -- IMO this is one of the oddest statements ever: "Linux has never been about ‘choice’ or ‘freedom’ and those myths should just die out. Read: https://www.redhat.com/archives/fedora-devel-list/2008-January/msg00861.html" - http://allanmcrae.com/2014/12/pacman-4-2-released/#comment-1256 Actually I use Linux (kernel and user space software) because it at least _was_ about manifoldness and libre (FLOSS). ^"L" is for libre
However, I like that Linus Torvalds made some clear statements to the careless, ignorant systemd and dbus crowed, while he's not against systemd and dbus, just against the new attitudes.
Writing good software is hard. Providing the developers of that software with concrete examples of use cases that `--asroot` provided may be useful to them. Providing patches may be useful. Asking for semantic versioning to be more strictly followed may be useful. Pointing out the value of backward compatibility (and then acknowledging that maintaining backward compatibility has a cost) may be useful. It is _not_ useful to then drag in systemd, dbus, Linus Torvalds, Red Hat, Johannes Löthberg, FLOSS philosophy and more. I'll re-read my copy of Hackers if I want to reminisce about the good old days of software development as it was in 1977 and enjoy a few good Ad Hominems. Please, let's aim for a constructive conversation. That's why I subscribe to this mailing list. —Jeremy
On Sat, 3 Jan 2015 18:57:11 -0500, Jeremy Audet wrote:
However, I like that Linus Torvalds made some clear statements to the careless, ignorant systemd and dbus crowed, while he's not against systemd and dbus, just against the new attitudes.
Writing good software is hard. Providing the developers of that software with concrete examples of use cases that `--asroot` provided may be useful to them. Providing patches may be useful. Asking for semantic versioning to be more strictly followed may be useful. Pointing out the value of backward compatibility (and then acknowledging that maintaining backward compatibility has a cost) may be useful. It is _not_ useful to then drag in systemd, dbus, Linus Torvalds, Red Hat, Johannes Löthberg, FLOSS philosophy and more. I'll re-read my copy of Hackers if I want to reminisce about the good old days of software development as it was in 1977 and enjoy a few good Ad Hominems.
Please, let's aim for a constructive conversation. That's why I subscribe to this mailing list.
The same for me. Once community was about computer people how did real life discussions and decisions. We tweaked hardware and software. The picture shows my Atari ST, modified for parallel usage of the build in 68000 + an additional 80286, the 512 KiB RAM are replaced by oldish 4 MiB PC RAM and you are free to chose between 2 MIDI in sockets [1]. I also could post pictures of at least one much older computer + hardware hacks and I might own or not own a lot of cracked software. JFTR the picture does show an original Steinberg dongle. We didn't crack the software to steal something! We pay for the software and then we needed to crack it, to fix issues, such as broken backwards compatibility ... to be fair Steinberg didn't suffer from this disease at this time. For me there was a good reason to switch to Linux and there also is a very good reason to communicate my opinion nowadays. I suspect computer nerd communities lost a lot when we got the Internet and dropped our soldering stations. All the QL, C64, Atari, Falcon etc. pp. nerds are still alive and much often decided to use Linux nowadays. Why should we agree with a policy of youngsters who don't care about compatibility and work-flows? [1] http://picpaste.de/pics/atari_st.1420331275.jpg Regards, Ralf
On Sat, Jan 3, 2015 at 4:34 PM, Ralf Mardorf <ralf.mardorf@rocketmail.com> wrote:
Linux and BSD user space is software based on other software. Yaourt and some other software does need pacman. Pacman needs bash, curl and other packages and those packages depend on other packages too.
This policy does only work, when there are agreements about backwards compatibility.
makepkg provided the --asroot option and within a major release it dropped backwards compatibility. Fortunately it is a minor issue that --asroot is dropped, but the real issue is that it already might be a fashion not to care about other software, not to care about work flows, IOW to be careless, ignorant.
So what exactly is the issue here? pacman has engaged in healthy thought about the respective pros/cons of removing --asroot, and you acknowledge that that is all right... therefore it necessitates mentioning that Just In Case on a completely unrelated note everyone gets attacked by a vicious case of wanting to drop backwards compatibility for the lulz, we should remember it is a Bad Thing to do so? -- Eli Schwartz
On 01/03/15 21:24, Eli Schwartz wrote:
On Sat, Jan 3, 2015 at 4:34 PM, Ralf Mardorf <ralf.mardorf@rocketmail.com> wrote:
Linux and BSD user space is software based on other software. Yaourt and some other software does need pacman. Pacman needs bash, curl and other packages and those packages depend on other packages too.
This policy does only work, when there are agreements about backwards compatibility.
makepkg provided the --asroot option and within a major release it dropped backwards compatibility. Fortunately it is a minor issue that --asroot is dropped, but the real issue is that it already might be a fashion not to care about other software, not to care about work flows, IOW to be careless, ignorant.
So what exactly is the issue here?
pacman has engaged in healthy thought about the respective pros/cons of removing --asroot, and you acknowledge that that is all right... therefore it necessitates mentioning that Just In Case on a completely unrelated note everyone gets attacked by a vicious case of wanting to drop backwards compatibility for the lulz, we should remember it is a Bad Thing to do so?
-- Eli Schwartz I believe the issue is deprecation of a feature utilized by an AUR (non officially sanctioned) package (yaourt). It should be the responsibility of the AUR maintainer to engage in discussions regarding deprecation of official features; and it should be the AUR maintainer's responsibility to modify their packages accordingly.
With that in mind; how will building on live cd's as non root work? Will users need to trivially make a pacman user for building packages? Regards, Mark
On Sat, Jan 3, 2015 at 9:11 PM, Mark Lee <mark@markelee.com> wrote:
I believe the issue is deprecation of a feature utilized by an AUR (non officially sanctioned) package (yaourt). It should be the responsibility of the AUR maintainer to engage in discussions regarding deprecation of official features; and it should be the AUR maintainer's responsibility to modify their packages accordingly.
With that in mind; how will building on live cd's as non root work? Will users need to trivially make a pacman user for building packages?
Regards, Mark
Please don't make things up, yaourt does not use --asroot unless you run yaourt as root, which means that whether you use yaourt is completely irrelevant. Running yaourt as root is going to make yaourt run makepkg as root, so the solution is... don't run yaourt as root? Since YAOURT ITSELF tries hard to discourage you from doing just that, I fail to see the problem. And even if it WAS a problem for yaourt, the idea that an aur package should or would restrict the further development of pacman and makepgk is absolutely ludicrous. Okay, that option can now be removed altogether. And in fact it was, because funnily enough, it seems the yaourt maintainer doesn't care one iota and happily added this change: https://github.com/archlinuxfr/yaourt/commit/3f60d14443ef14687e4e2cb60c92209... which leaves yaourt brilliantly happy with pacman 4.2 In short, this absolutely not in any way shape or form about aur helpers. It is about people who want to build packages (however they care to do so) from a root shell. -- Eli Schwartz
On 01/03/15 23:18, Eli Schwartz wrote:
On Sat, Jan 3, 2015 at 9:11 PM, Mark Lee <mark@markelee.com> wrote:
I believe the issue is deprecation of a feature utilized by an AUR (non officially sanctioned) package (yaourt). It should be the responsibility of the AUR maintainer to engage in discussions regarding deprecation of official features; and it should be the AUR maintainer's responsibility to modify their packages accordingly.
With that in mind; how will building on live cd's as non root work? Will users need to trivially make a pacman user for building packages?
Regards, Mark
Please don't make things up, yaourt does not use --asroot unless you run yaourt as root, which means that whether you use yaourt is completely irrelevant.
Running yaourt as root is going to make yaourt run makepkg as root, so the solution is... don't run yaourt as root? Since YAOURT ITSELF tries hard to discourage you from doing just that, I fail to see the problem.
And even if it WAS a problem for yaourt, the idea that an aur package should or would restrict the further development of pacman and makepgk is absolutely ludicrous.
Okay, that option can now be removed altogether. And in fact it was, because funnily enough, it seems the yaourt maintainer doesn't care one iota and happily added this change: https://github.com/archlinuxfr/yaourt/commit/3f60d14443ef14687e4e2cb60c92209... which leaves yaourt brilliantly happy with pacman 4.2
In short, this absolutely not in any way shape or form about aur helpers. It is about people who want to build packages (however they care to do so) from a root shell.
-- Eli Schwartz An AUR (unofficial package) shouldn't be restricting the development of an official package. But, a dialogue regarding any differences surely can take place.
If you think it's solely about building from a root shell; you should re-read the thread. It surely has progressed past that dialogue. Calm down for a bit and don't miss the forrest for the trees. Regards, Mark
On Sat, Jan 3, 2015 at 10:23 PM, Mark Lee <mark@markelee.com> wrote:
An AUR (unofficial package) shouldn't be restricting the development of an official package. But, a dialogue regarding any differences surely can take place.
No dialog is necessary, it was a straightforward change with absolutely no impact beyond the impact of running makepkg --asroot by hand instead of by proxy. In fact, the only change to yaourt is making yaourt exit when run as root instead of warning you it's dangerous. If you think it's solely about building from a root shell; you should
re-read the thread. It surely has progressed past that dialogue.
Please elaborate about these additional concerns. I see no additional concerns, it is all about people who would like to build from a root shell. Whether they should or shouldn't, I don't concern myself, but that is the issue at hand. Considering that the actual discussion is about makepkg no longer allowing you to run from a root shell, that is kind of natural. (I do not count offtopic discussion as being a concern about this change.) And as I said in the first place, you are factually incorrect in stating that the concern is with aur packages. That is FUD originally posted in the comments on Allan McRae's blog[1] about which the only mention here was that it too is factually incorrect. And even if aur packages were broken (it seems there are indeed aur helpers that run as root) that is still not pacman's fault in this particular case, that is a bug that bust be fixed regardless in said aur helper. [1] http://allanmcrae.com/2014/12/pacman-4-2-released/#comment-1250
Calm down for a bit and don't miss the forrest for the trees.
Regards, Mark
I see no forest, merely a lonesome tree on a deserted plain. -- Eli Schwartz
On 01/04/15 00:09, Eli Schwartz wrote:
On Sat, Jan 3, 2015 at 10:23 PM, Mark Lee <mark@markelee.com> wrote:
An AUR (unofficial package) shouldn't be restricting the development of an official package. But, a dialogue regarding any differences surely can take place.
No dialog is necessary, it was a straightforward change with absolutely no impact beyond the impact of running makepkg --asroot by hand instead of by proxy.
In fact, the only change to yaourt is making yaourt exit when run as root instead of warning you it's dangerous.
If you think it's solely about building from a root shell; you should
re-read the thread. It surely has progressed past that dialogue.
Please elaborate about these additional concerns. I see no additional concerns, it is all about people who would like to build from a root shell. Whether they should or shouldn't, I don't concern myself, but that is the issue at hand. Considering that the actual discussion is about makepkg no longer allowing you to run from a root shell, that is kind of natural.
(I do not count offtopic discussion as being a concern about this change.)
And as I said in the first place, you are factually incorrect in stating that the concern is with aur packages. That is FUD originally posted in the comments on Allan McRae's blog[1] about which the only mention here was that it too is factually incorrect. And even if aur packages were broken (it seems there are indeed aur helpers that run as root) that is still not pacman's fault in this particular case, that is a bug that bust be fixed regardless in said aur helper.
[1] http://allanmcrae.com/2014/12/pacman-4-2-released/#comment-1250
Calm down for a bit and don't miss the forrest for the trees.
Regards, Mark
I see no forest, merely a lonesome tree on a deserted plain.
-- Eli Schwartz DId you read any of Ralf's messages? He prefaces by commenting about backwards compatibility and Torvalds' comments.
This isn't just about makepkg; this is about backwards compatibility as well. Just because yaourt changed it's process to match upstream doesn't change any of the previous concerns about backwards compatibility and dialogue regarding changes. Regards, Mark
On Sat, Jan 3, 2015 at 11:18 PM, Mark Lee <mark@markelee.com> wrote:
DId you read any of Ralf's messages? He prefaces by commenting about backwards compatibility and Torvalds' comments.
This isn't just about makepkg; this is about backwards compatibility as well. Just because yaourt changed it's process to match upstream doesn't change any of the previous concerns about backwards compatibility and dialogue regarding changes.
Regards, Mark
That was the offtopic I referred to. Well, some of it. Since it doesn't actually have anything to do with this change. As he himself stated: On Sat, Jan 3, 2015 at 4:34 PM, Ralf Mardorf <ralf.mardorf@rocketmail.com> wrote:
Linux and BSD user space is software based on other software. Yaourt and some other software does need pacman. Pacman needs bash, curl and other packages and those packages depend on other packages too.
This policy does only work, when there are agreements about backwards compatibility.
makepkg provided the --asroot option and within a major release it dropped backwards compatibility. Fortunately it is a minor issue that --asroot is dropped, but the real issue is that it already might be a fashion not to care about other software, not to care about work flows, IOW to be careless, ignorant.
Linus and systemd, while no doubt exciting, doesn't mean that the ArchLinux developers have decided it is cool to break backwards compatibility. So while have no intention of arguing with the concern for backwards compatibility, pacman 4.2 and the removal of --asroot is not a backwards-compatibility issue in the end, as the very person who brought it up ultimately determined. And thus, the concern about backwards compatibility is offtopic in this thread -- Eli Schwartz
On 01/04/15 00:40, Eli Schwartz wrote:
On Sat, Jan 3, 2015 at 11:18 PM, Mark Lee <mark@markelee.com> wrote:
DId you read any of Ralf's messages? He prefaces by commenting about backwards compatibility and Torvalds' comments.
This isn't just about makepkg; this is about backwards compatibility as well. Just because yaourt changed it's process to match upstream doesn't change any of the previous concerns about backwards compatibility and dialogue regarding changes.
Regards, Mark
That was the offtopic I referred to. Well, some of it.
Since it doesn't actually have anything to do with this change.
As he himself stated:
On Sat, Jan 3, 2015 at 4:34 PM, Ralf Mardorf <ralf.mardorf@rocketmail.com> wrote:
Linux and BSD user space is software based on other software. Yaourt and some other software does need pacman. Pacman needs bash, curl and other packages and those packages depend on other packages too.
This policy does only work, when there are agreements about backwards compatibility.
makepkg provided the --asroot option and within a major release it dropped backwards compatibility. Fortunately it is a minor issue that --asroot is dropped, but the real issue is that it already might be a fashion not to care about other software, not to care about work flows, IOW to be careless, ignorant.
Linus and systemd, while no doubt exciting, doesn't mean that the ArchLinux developers have decided it is cool to break backwards compatibility. So while have no intention of arguing with the concern for backwards compatibility, pacman 4.2 and the removal of --asroot is not a backwards-compatibility issue in the end, as the very person who brought it up ultimately determined.
And thus, the concern about backwards compatibility is offtopic in this thread
-- Eli Schwartz I concur that the title is incorrect.
Regards, Mark
On Sun, 04 Jan 2015 00:18:25 -0500 Mark Lee <mark@markelee.com> wrote:
No dialog is necessary, it was a straightforward change with absolutely no impact
That statement is the Absolute height of obnoxious arrogant ignorance ans hows your thinking for what it is .. i am so i am right attitude .. disgusting Pete . -- Illegitimi non carborundum . ro for the purists out there Noli nothis permittere te terere.
On Sat, 3 Jan 2015 23:09:01 -0600 Eli Schwartz <eschwartz93@gmail.com> wrote:
No dialog is necessary, it was a straightforward change with absolutely no impact
Sent to the correct header this time . Apologies Mark That statement is the Absolute height of obnoxious arrogant ignorance ans hows your thinking for what it is .. i am so i am right attitude .. disgusting Pete . -- Illegitimi non carborundum . ro for the purists out there Noli nothis permittere te terere.
On Sun, Jan 4, 2015 at 1:59 AM, pete nikolic <pg.nikolic1@gmail.com> wrote:
On Sat, 3 Jan 2015 23:09:01 -0600 Eli Schwartz <eschwartz93@gmail.com> wrote:
No dialog is necessary, it was a straightforward change with absolutely no impact
Sent to the correct header this time .
Apologies Mark
That statement is the Absolute height of obnoxious arrogant ignorance ans hows your thinking for what it is .. i am so i am right attitude .. disgusting
Pete .
Well, you could at least actually quote what I really said: "[...] it was a straightforward change with absolutely no impact" ... now wait for it :) ... "beyond the impact of running makepkg --asroot by hand instead of by proxy." I was, of course, referring solely to the nature of that change as it regards yaourt and the yaourt maintainer needing to discuss the change with the devs. Which is to say, this change does not affect yaourt in any appreciable way, and the people it does affect... will be affected without bringing yaourt into the discussion. I have absolutely no opinion on people wishing to run "makepkg --asroot", that is absolutely your right and I wish you much luck in finding a suitable workaround or convincing the pacman developers of its merits or whatever you like. Because I have no opinion on whether or not it is a legitimate need, or when it might or might not be a need with extra legitimacy behind it... beyond observing that I have never used it, myself, personally. -- Eli Schwartz
On Sun, 4 Jan 2015 18:09:19 -0600, Eli Schwartz wrote:
Well, you could at least actually quote what I really said:
"[...] it was a straightforward change with absolutely no impact" ... now wait for it :) ... "beyond the impact of running makepkg --asroot by hand instead of by proxy."
I was, of course, referring solely to the nature of that change as it regards yaourt and the yaourt maintainer needing to discuss the change with the devs. Which is to say, this change does not affect yaourt in any appreciable way, and the people it does affect... will be affected without bringing yaourt into the discussion.
I have absolutely no opinion on people wishing to run "makepkg --asroot", that is absolutely your right and I wish you much luck in finding a suitable workaround or convincing the pacman developers of its merits or whatever you like. Because I have no opinion on whether or not it is a legitimate need, or when it might or might not be a need with extra legitimacy behind it... beyond observing that I have never used it, myself, personally.
For my taste pacman is the best package management I know. It still is, after a makepkg option was dropped. I'm a yaourt user and can't notice any difference, since I usually run yaourt without root privileges. Sometimes I run makepkg as root, but it's no issue for me to run it without root privileges, it indeed is smarter to do it that way. The attitude to ignore the Linux ecosystem all in all, that became a fashion a while ago is disgusting. The way to diss software contributed by others, the way to disgrade other users who might have less knowledge, who might belong to a minority is outrageous. Reducing "community" to people who contribute to Linux is alarming. FLOSS is based on public-spirited world views. Dropping backwards compatibility within a major release (for a ".dot" release) is ignorant. The tone of voice and no willingness to care about others, including minorities is embarrassing.
On Mon, 5 Jan 2015 01:47:05 +0100 Ralf Mardorf <ralf.mardorf@rocketmail.com> wrote:
On Sun, 4 Jan 2015 18:09:19 -0600, Eli Schwartz wrote:
Well, you could at least actually quote what I really said:
"[...] it was a straightforward change with absolutely no impact" ... now wait for it :) ... "beyond the impact of running makepkg --asroot by hand instead of by proxy."
I was, of course, referring solely to the nature of that change as it regards yaourt and the yaourt maintainer needing to discuss the change with the devs. Which is to say, this change does not affect yaourt in any appreciable way, and the people it does affect... will be affected without bringing yaourt into the discussion.
I have absolutely no opinion on people wishing to run "makepkg --asroot", that is absolutely your right and I wish you much luck in finding a suitable workaround or convincing the pacman developers of its merits or whatever you like. Because I have no opinion on whether or not it is a legitimate need, or when it might or might not be a need with extra legitimacy behind it... beyond observing that I have never used it, myself, personally.
For my taste pacman is the best package management I know. It still is, after a makepkg option was dropped. I'm a yaourt user and can't notice any difference, since I usually run yaourt without root privileges. Sometimes I run makepkg as root, but it's no issue for me to run it without root privileges, it indeed is smarter to do it that way.
The attitude to ignore the Linux ecosystem all in all, that became a fashion a while ago is disgusting. The way to diss software contributed by others, the way to disgrade other users who might have less knowledge, who might belong to a minority is outrageous. Reducing "community" to people who contribute to Linux is alarming. FLOSS is based on public-spirited world views. Dropping backwards compatibility within a major release (for a ".dot" release) is ignorant. The tone of voice and no willingness to care about others, including minorities is embarrassing.
How in the world do "minorities" have anything to do with anything in this thread?
On Sun, 4 Jan 2015 19:14:02 -0600, Doug Newgard wrote:
How in the world do "minorities" have anything to do with anything in this thread?
A few people mentioned that they run into issues. Did you miss their messages? For example: On Sun, 04 Jan 2015 19:48:14 +0100, Marcel Kleinfeller wrote: [snip]
[marcel@oompf ~]$ sudo packer -Syu [snip] makepkg: Ungültige Option '--asroot' The build failed [snip]
The OP should have used LANG=C or LC_ALL=C to get rid of the German messages.
On Mon, 5 Jan 2015 02:32:58 +0100 Ralf Mardorf <ralf.mardorf@rocketmail.com> wrote:
On Sun, 4 Jan 2015 19:14:02 -0600, Doug Newgard wrote:
How in the world do "minorities" have anything to do with anything in this thread?
A few people mentioned that they run into issues. Did you miss their messages?
For example: On Sun, 04 Jan 2015 19:48:14 +0100, Marcel Kleinfeller wrote: [snip]
[marcel@oompf ~]$ sudo packer -Syu [snip] makepkg: Ungültige Option '--asroot' The build failed [snip]
The OP should have used LANG=C or LC_ALL=C to get rid of the German messages.
No, I did not miss them, your terminology was just strange. The people that have issues need to be educated to do it differently, nothing more.
On Sun, 4 Jan 2015 19:45:10 -0600, Doug Newgard wrote:
On Mon, 5 Jan 2015 02:32:58 +0100 Ralf Mardorf <ralf.mardorf@rocketmail.com> wrote:
On Sun, 4 Jan 2015 19:14:02 -0600, Doug Newgard wrote:
How in the world do "minorities" have anything to do with anything in this thread?
A few people mentioned that they run into issues. Did you miss their messages?
For example: On Sun, 04 Jan 2015 19:48:14 +0100, Marcel Kleinfeller wrote: [snip]
[marcel@oompf ~]$ sudo packer -Syu [snip] makepkg: Ungültige Option '--asroot' The build failed [snip]
The OP should have used LANG=C or LC_ALL=C to get rid of the German messages.
No, I did not miss them, your terminology was just strange. The people that have issues need to be educated to do it differently, nothing more. ^^^^^^^^ Do they need a teacher? The reasons to drop '--asroot' are valid reasons. The style it was dropped IMO is a problem. Fakeroot is a Debian thingy. What does happen if the fakeroot developers ignore common sense of backwards compatibility within a major release? ;)
Fakeroot depends on sed. What happens assumed an option for sed should be dropped within a major release? Is community spirit out of style?
On Mon, 5 Jan 2015 02:56:31 +0100 Ralf Mardorf <ralf.mardorf@rocketmail.com> wrote:
On Sun, 4 Jan 2015 19:45:10 -0600, Doug Newgard wrote:
On Mon, 5 Jan 2015 02:32:58 +0100 Ralf Mardorf <ralf.mardorf@rocketmail.com> wrote:
On Sun, 4 Jan 2015 19:14:02 -0600, Doug Newgard wrote:
How in the world do "minorities" have anything to do with anything in this thread?
A few people mentioned that they run into issues. Did you miss their messages?
For example: On Sun, 04 Jan 2015 19:48:14 +0100, Marcel Kleinfeller wrote: [snip]
[marcel@oompf ~]$ sudo packer -Syu [snip] makepkg: Ungültige Option '--asroot' The build failed [snip]
The OP should have used LANG=C or LC_ALL=C to get rid of the German messages.
No, I did not miss them, your terminology was just strange. The people that have issues need to be educated to do it differently, nothing more. ^^^^^^^^ Do they need a teacher? The reasons to drop '--asroot' are valid reasons. The style it was dropped IMO is a problem. Fakeroot is a Debian thingy. What does happen if the fakeroot developers ignore common sense of backwards compatibility within a major release? ;)
Fakeroot depends on sed. What happens assumed an option for sed should be dropped within a major release?
Is community spirit out of style?
Yeah, it's not like sed options are defined by POSIX or anything... Your entire problem here seems to be that you don't like the development style of pacman. In that case, get involved instead of just whining about it after the fact. It's been well known that this has been coming for over 9 months. And for the record, I have see problem making relatively minor changes like this with a minor release. If you think it's a problem, you're going to need a better argument that "it's common sense". What you see as common sense, I see as arbitrary and artificial restrictions for no good reason.
On Sun, Jan 4, 2015 at 8:16 PM, Doug Newgard <scimmia@archlinux.info> wrote:
Yeah, it's not like sed options are defined by POSIX or anything...
Your entire problem here seems to be that you don't like the development style of pacman. In that case, get involved instead of just whining about it after the fact. It's been well known that this has been coming for over 9 months.
And for the record, I have see problem making relatively minor changes like this with a minor release. If you think it's a problem, you're going to need a better argument that "it's common sense". What you see as common sense, I see as arbitrary and artificial restrictions for no good reason.
I myself would dearly love to know what precisely about removing a discouraged option qualifies as a major change. -- Eli Schwartz
On Sun, 4 Jan 2015 20:25:11 -0600, Eli Schwartz wrote:
I myself would dearly love to know what precisely about removing a discouraged option qualifies as a major change.
Maybe the answer depends to the reason, that the discouraged option was provided for some _unknown_ reason. I absolutely agree that dropping --asroot is something we can accept. But I'm likely not the only one who noticed that backwards compatibility became a serious issue for Linux user space software. I mentioned Torvalds, because even he is unable to make clear why some policies are completely wrong. At least he has got the decision-making power not to implement everything into the kernel. If those guys who quote Redhad sources as the one and only truth would have more decision-making power, than Torvalds has got, our kernels would be buggy Redhad ruled things, with broken dbus implementation and other damages. Fortunately the kernel is protected, but user space isn't protect, so IMO we should be noisy when things happen that might be borderline. -- IMO this is one of the oddest statements ever: "Linux has never been about ‘choice’ or ‘freedom’ and those myths should just die out. Read: https://www.redhat.com/archives/fedora-devel-list/2008-January/msg00861.html" - http://allanmcrae.com/2014/12/pacman-4-2-released/#comment-1256 Actually I use Linux (kernel and user space software) because it at least _was_ about manifoldness and libre (FLOSS). ^"L" is for libre
On Mon, 5 Jan 2015 03:55:23 +0100 Ralf Mardorf <ralf.mardorf@rocketmail.com> wrote:
On Sun, 4 Jan 2015 20:25:11 -0600, Eli Schwartz wrote:
I myself would dearly love to know what precisely about removing a discouraged option qualifies as a major change.
Maybe the answer depends to the reason, that the discouraged option was provided for some _unknown_ reason. ...
The reason is mentioned in the commit removing the option: https://projects.archlinux.org/pacman.git/commit/?id=61ba5c961e4a3536c4bbf41...
On Sun, Jan 4, 2015 at 8:55 PM, Ralf Mardorf <ralf.mardorf@rocketmail.com> wrote:
Maybe the answer depends to the reason, that the discouraged option was provided for some _unknown_ reason. I absolutely agree that dropping --asroot is something we can accept. But I'm likely not the only one who noticed that backwards compatibility became a serious issue for Linux user space software. I mentioned Torvalds, because even he is unable to make clear why some policies are completely wrong. At least he has got the decision-making power not to implement everything into the kernel. If those guys who quote Redhad sources as the one and only truth would have more decision-making power, than Torvalds has got, our kernels would be buggy Redhad ruled things, with broken dbus implementation and other damages. Fortunately the kernel is protected, but user space isn't protect, so IMO we should be noisy when things happen that might be borderline.
Oh, so you admit that you are riding your hobby horse here, and that it isn't really about pacman at all, but rather about a completely unrelated topic, e.g. What Is Wrong With The Linux Community These Days? In short, and this is what I have been wondering for most of this thread, why are we using the Arch-general mailing list to discuss our personal disappointments with Redhat development policy, by pretending we are talking about pacman, when really we aren't actually talking about pacman? pacman 4.2 is not indicative of the same mindset as Redhat/Poettering/systemd/dbus/whatever, and there is no need to artificially tie the two together merely to gain a soapbox upon which to stand and declaim against Redhat. Get a blog. -- Eli Schwartz
On Sun, Jan 4, 2015 at 6:47 PM, Ralf Mardorf <ralf.mardorf@rocketmail.com> wrote:
The attitude to ignore the Linux ecosystem all in all, that became a fashion a while ago is disgusting. The way to diss software contributed by others, the way to disgrade other users who might have less knowledge, who might belong to a minority is outrageous. Reducing "community" to people who contribute to Linux is alarming. FLOSS is based on public-spirited world views. Dropping backwards compatibility within a major release (for a ".dot" release) is ignorant. The tone of voice and no willingness to care about others, including minorities is embarrassing.
I am content to not-care as much as I please. This is a workflow issue, I am not obligated to get on my hobby horse over it, unless someone is actually physically unable to adjust their workflow. (Like the guy compiling for a chroot on android, who found an acceptable workaround and incidentally was very polite about the whole thing despite, unusually enough, having a legitimate problem.) -- Eli Schwartz
participants (12)
-
Christian Demsar
-
Doug Newgard
-
Eli Schwartz
-
Ian D. Scott
-
Jeremy Audet
-
Leonid Isaev
-
Magnus Therning
-
Mark Lee
-
member graysky
-
pete nikolic
-
Ralf Mardorf
-
Yamakaky