Bug#776557: debian-policy: Please clarify 2.5 'unix heritage >= important'
> Le Thu, Jan 29, 2015 at 09:06:15AM +, Martin Zobel-Helas a écrit : > > > > the following sentence in 2.5 leave much room for maneuver, therefor i > > would like to see a clarification how it should be interpreted: > > > > | Important programs, including those which one would expect to find on > > | any Unix-like system. If the expectation is that an experienced Unix > > | person who found it missing would say "What on earth is going on, where > > | is foo?", it must be an important package. Le Thu, Jan 29, 2015 at 10:03:27PM +0900, Charles Plessy a écrit : > > Given that Debian is 20 years old, we can not expect people to have the same > opinion on "What on earth is going on, where is foo?" means. On my side, I > thought that "killall" or "less" would be "what-on-earth" programs, but this > is > not the case. My first reaction was to argue they should be present by > default > on minimal systems, but my current opinion would be to rather keep minimal > systems as lean as possible and rely on tasks for adding groups of packages. > > Regarding the Policy, we need to either find a different principle for > defining > the "Important" priority, or transfer the responsibility for choices to a > "do-o-cratic" group of persons, like people making minimal images, maintaining > debootstrap, etc. (and by default, the package maintainer of course) Hello everybody, stimulated by the progesses in #758234, I would like to propose a new definition for the "important" Priority. First, let's look at the the definition of "required". It is very straigthforward: the bare minimum needed to run dpkg. Interstingly, after a quick look at the list of "important" packages, I have the impression that they are close to the minimum needed to run apt over the network. If you agree with my analysis, I think that the Policy would be clearer with the following alternative definition for "important". (The last sentence is there because man-db, debian-faq and locales are all priority:standard.) Packages which are necessary for a system to run `apt` and use it to download other packages from the network, plus the bare minimum of commonly-expected and necessary tools to administrate that system. This does not include space-consuming features such as documentation and multilingual support. Have a nice day, (Please CC me, I have not yet resubscribed to the list) -- Charles Plessy Tsurumi, Kanagawa, Japan
Bug#776557: debian-policy: Please clarify 2.5 'unix heritage = important'
On Thu, Jan 29, 2015 at 09:06:15AM +, Martin Zobel-Helas wrote: Package: debian-policy Severity: important Hi, the following sentence in 2.5 leave much room for maneuver, therefor i would like to see a clarification how it should be interpreted: | Important programs, including those which one would expect to find on | any Unix-like system. If the expectation is that an experienced Unix | person who found it missing would say What on earth is going on, where | is foo?, it must be an important package. Background here is, that i moved the package ed to optional years ago, and now have bug #776413 open, which disagrees on that move. I would like to keep ed in optional, but also see the arguments the submitter gave here. You rightfull object that 'the expectation is that an experienced Unix person' is subjective in nature and leads to contradictory opinion. So maybe we should replace the phrase 'If the expectation is ...' by a reference to standard that define what UNIX is (POSIX, SUS). However, as far as I understand, ed is mandated by POSIX and SUSv4. So I do not see how we can keep important to refer to UNIX and at the same time excludind ed. Cheers, -- Bill. ballo...@debian.org Imagine a large red swirl here. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#776557: debian-policy: Please clarify 2.5 'unix heritage = important'
Bill Allombert ballo...@debian.org writes: You rightfull object that 'the expectation is that an experienced Unix person' is subjective in nature and leads to contradictory opinion. So maybe we should replace the phrase 'If the expectation is ...' by a reference to standard that define what UNIX is (POSIX, SUS). However, as far as I understand, ed is mandated by POSIX and SUSv4. So I do not see how we can keep important to refer to UNIX and at the same time excluding ed. Well, the current Policy rule is intentionally subjective. The idea is to focus on the user experience, not on a particular standard. Standards always have a problem: once you put something in a standard, it's nearly impossible to take it out again. POSIX is pretty unlikely to ever remove ed, even if no one uses it, since there's no way to know what's using it or what vendor scripts, etc., have built in some assumption that it is available. And, more practically, because it would require a lot of work to achieve consensus on removing it, and people would object just because people always object to this sort of thing in standards bodies, so no one will bother to do the work of getting it removed. I think the question here is whether we want to treat issues like this the same way, or whether we want to use some other standard of general usefulness for things in important. Or, put another way, how much weight do we want to put on standards and on the fact that something has historically been in important? If we were building the important set from scratch based on things that are commonly needed for a minimal system, I doubt we would include ed. I cannot remember the last time I ran into something that actually uses it. But we're not, and the package isn't all that large. Personally, I would tend to lean towards letting the people who work on the installer and the CD sets and similar space-constrained areas of Debian (embedded environments, maybe, as well) be the ones who decide the membership of standard vs. important vs. required rather than having each individual maintainer roll the dice and apply their own personal guesswork. -- Russ Allbery (r...@debian.org) http://www.eyrie.org/~eagle/ -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#776557: debian-policy: Please clarify 2.5 'unix heritage = important'
Hi, Bill Allombert: On the other hand, ed scripts are still in use: - by diff Reasonable people have been using diff -u for the last umpteen years. - by apt-get: the pdiff system use ed scripts which I assume has a dependency on ed. so it is useful to keep the reference to what this is all about somewhere. Sure. Nobody wants to drop ed from Debian. (I hope.) -- -- Matthias Urlichs -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#776557: debian-policy: Please clarify 2.5 'unix heritage = important'
On Thu, Jan 29, 2015 at 10:49:42AM +0100, Matthias Urlichs wrote: Hi, Martin Zobel-Helas: the following sentence in 2.5 leave much room for maneuver, therefor i would like to see a clarification how it should be interpreted: | Important programs, including those which one would expect to find on | any Unix-like system. If the expectation is that an experienced Unix | person who found it missing would say What on earth is going on, where | is foo?, it must be an important package. Background here is, that i moved the package ed to optional years ago, and now have bug #776413 open, which disagrees on that move. Quite frankly, any experienced Unix person who still uses ed for, well, anything at all really, should ask themselves where the hell they've been during the last 20 years or so. I do not think people are still using ed for interactive use (unless they have no other choice left) On the other hand, ed scripts are still in use: - by diff - by apt-get: the pdiff system use ed scripts so it is useful to keep the reference to what this is all about somewhere. Cheers, -- Bill. ballo...@debian.org Imagine a large red swirl here. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#776557: debian-policy: Please clarify 2.5 'unix heritage = important'
On 2015-01-29 16:55, Matthias Urlichs wrote: Hi, Bill Allombert: [...] - by apt-get: the pdiff system use ed scripts which I assume has a dependency on ed. apt uses an internal implementation. Regards, Adam -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#776557: debian-policy: Please clarify 2.5 'unix heritage = important'
Package: debian-policy Severity: important Hi, the following sentence in 2.5 leave much room for maneuver, therefor i would like to see a clarification how it should be interpreted: | Important programs, including those which one would expect to find on | any Unix-like system. If the expectation is that an experienced Unix | person who found it missing would say What on earth is going on, where | is foo?, it must be an important package. Background here is, that i moved the package ed to optional years ago, and now have bug #776413 open, which disagrees on that move. I would like to keep ed in optional, but also see the arguments the submitter gave here. Hoping for your assistance, Martin -- System Information: Debian Release: 8.0 APT prefers testing-updates APT policy: (500, 'testing-updates'), (500, 'testing'), (1, 'experimental') Architecture: amd64 (x86_64) Foreign Architectures: i386 Kernel: Linux 3.18.0-trunk-amd64 (SMP w/8 CPU cores) Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8) Shell: /bin/sh linked to /bin/dash Init: systemd (via /run/systemd/system) -- Martin Zobel-Helas zo...@debian.orgDebian System Administrator Debian GNU/Linux Developer Debian Listmaster http://about.me/zobel Debian Webmaster GPG Fingerprint: 6B18 5642 8E41 EC89 3D5D BDBB 53B1 AC6D B11B 627B signature.asc Description: Digital signature
Bug#776557: debian-policy: Please clarify 2.5 'unix heritage = important'
Hi, Martin Zobel-Helas: the following sentence in 2.5 leave much room for maneuver, therefor i would like to see a clarification how it should be interpreted: | Important programs, including those which one would expect to find on | any Unix-like system. If the expectation is that an experienced Unix | person who found it missing would say What on earth is going on, where | is foo?, it must be an important package. Background here is, that i moved the package ed to optional years ago, and now have bug #776413 open, which disagrees on that move. Quite frankly, any experienced Unix person who still uses ed for, well, anything at all really, should ask themselves where the hell they've been during the last 20 years or so. -- -- Matthias Urlichs -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#776557: debian-policy: Please clarify 2.5 'unix heritage = important'
Le Thu, Jan 29, 2015 at 09:06:15AM +, Martin Zobel-Helas a écrit : the following sentence in 2.5 leave much room for maneuver, therefor i would like to see a clarification how it should be interpreted: | Important programs, including those which one would expect to find on | any Unix-like system. If the expectation is that an experienced Unix | person who found it missing would say What on earth is going on, where | is foo?, it must be an important package. Background here is, that i moved the package ed to optional years ago, and now have bug #776413 open, which disagrees on that move. I would like to keep ed in optional, but also see the arguments the submitter gave here. Hi Martin, I fully agree. Given that Debian is 20 years old, we can not expect people to have the same opinion on What on earth is going on, where is foo? means. On my side, I thought that killall or less would be what-on-earth programs, but this is not the case. My first reaction was to argue they should be present by default on minimal systems, but my current opinion would be to rather keep minimal systems as lean as possible and rely on tasks for adding groups of packages. Regarding the Policy, we need to either find a different principle for defining the Important priority, or transfer the responsibility for choices to a do-o-cratic group of persons, like people making minimal images, maintaining debootstrap, etc. (and by default, the package maintainer of course) Have a nice day, -- Charles Plessy Debian Med packaging team, http://www.debian.org/devel/debian-med Tsurumi, Kanagawa, Japan -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#776557: debian-policy: Please clarify 2.5 'unix heritage = important'
On Thu, Jan 29, 2015 at 05:55:11PM +0100, Matthias Urlichs wrote: Hi, Bill Allombert: On the other hand, ed scripts are still in use: - by diff Reasonable people have been using diff -u for the last umpteen years. Why so much drama ? I am sure you did not intend to call me unreasonnable, yet your post suggests it. It is unfortunate it is not possible for diff to default to -u (or even -c) without breaking backward compatility (including POSIX compatility). Cheers, -- Bill. ballo...@debian.org Imagine a large red swirl here. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#776557: debian-policy: Please clarify 2.5 'unix heritage = important'
Hi, Bill Allombert: On the other hand, ed scripts are still in use: - by diff Reasonable people have been using diff -u for the last umpteen years. Why so much drama ? I am sure you did not intend to call me unreasonnable, yet your post suggests it. No, of course that was not intended. The intent was merely to state that ed-as-an-interpreter-for-diff is *way* obsolete. It is unfortunate it is not possible for diff to default to -u (or even -c) without breaking backward compatility (including POSIX compatility). Right. :-/ -- -- Matthias Urlichs -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#776557: debian-policy: Please clarify 2.5 'unix heritage = important'
Hi, On Thu Jan 29, 2015 at 21:36:11 +0100, Matthias Urlichs wrote: Hi, Bill Allombert: On the other hand, ed scripts are still in use: - by diff Reasonable people have been using diff -u for the last umpteen years. Why so much drama ? I am sure you did not intend to call me unreasonnable, yet your post suggests it. No, of course that was not intended. The intent was merely to state that ed-as-an-interpreter-for-diff is *way* obsolete. It is unfortunate it is not possible for diff to default to -u (or even -c) without breaking backward compatility (including POSIX compatility). Right. :-/ can we get back to my question of either clarifying 2.5 or updating that text? I find the definition of What on earth is going on, where is foo? not very helpful to define if a package is important or not. Cheers, Martin -- Martin Zobel-Helas zo...@debian.orgDebian System Administrator Debian GNU/Linux Developer Debian Listmaster http://about.me/zobel Debian Webmaster GPG Fingerprint: 6B18 5642 8E41 EC89 3D5D BDBB 53B1 AC6D B11B 627B -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org