We may not have had an opinion from Richard Stallman on systemd so far, but Linus gave his again in an interview with ITWire's Sam Varghese, where he also talks about other things.
When it comes to systemd, they expect me to have a lot of colorful opinions, but no. I don't personally care about systemd, in fact my main computer and laptop use it. Now, I don't get along with some of the developers (referring to the Kay Sievers incident) and I think they are a little carefree when it comes to bugs and compatibility, but I'm not much in the camp of people who hate the idea of systemd either. .
Do you agree with the thought that systemd moves away from the idea of simplicity of UNIX systems? That's bad?.
I believe that many of the "original ideas" of UNIX are more a matter of mindset than a reflection of reality. There is still value in understanding the traditional "do one thing and do it well" model, but that is not how complex systems work, and it is not how large applications were designed for a long time. This is a useful simplification that is true on a "certain" level, but clearly does not describe most of reality. And systemd is by no means the piece that breaks the old legacy UNIX. Graphical applications rarely work like this, and then there is obviously the traditional GNU Emacs counterexample, which was never just a simple UNIX model, but a big new infrastructure, like systemd. Of course, I am old enough to like logs in text and not in binary. Sometimes I think systemd doesn't necessarily have the best taste, but they are details.
Have you been through similar situations before, where the introduction of a new way of doing things caused so much bitterness and extreme reactions?
Oh yeah. Vi vs Emacs, the comparisons between desktops or for a closer case to systemd, SysV init vs BSD init. I'm not sure how different systemd fights are. It's a technical question, but its developers were certainly very good at alienating people on a personal level. It is not something new under the sun, the wars between those who defend the GPL and the BSD were more about the people involved and how they irritated people than other differences.
What would you say if someone argued that systemd created a failure point which does not allow the system to boot if it fails? It centralizes many services and if one fails the system is useless.
If that's a reason, they shouldn't use the kernel either. Obviously it is something special, that its engineers are better people and that perhaps it would be unfair to compare something as noble as the kernel with more mundane projects. But if you look at slow and heavy projects like glibc, when they screw up they all get hurt.
I asked this because I saw articles of people moving to BSD on servers. I hadn't seen such extreme behavior, but I've only been on Linux since '98.
I don't usually follow those storms, but I also think that one thing that's changing is that people are perhaps taking that culture of sharp, populist panic very seriously. It is not only in the technical press but in the technical world there are also many "opinion pieces" and other related exaggerations. And the BSD folks have a term for that: 'bikeshed painting' *, which is about how random people feel they have the ability to discuss superficial issues, because everyone feels they can give an opinion on the choice of color. So superficial issues have more noise. But when it comes to difficult and deeply technical decisions, they find (sometimes) that they don't know enough and that's why they don't talk as much.
You've read Lennart Poettering's new document about arranging distributions with the default Btrfs file system? If so, what do you think?
I'm not so sure it's necessarily the right way to do things, but I'm actually very happy that people are working in that direction. The current packaging model is broken for third-party applications, and I am not convinced that it is that good even for projects that are distributed on Linux distributions as part of their core. Are the exact details of how to use Btrfs to implement this the correct thing to do? No idea. It is a complex problem that is not going to be solved overnight with some radical novelty, and I am suspicious of new sophisticated models that change everything and say they solve problems (perhaps the novelty, the complexity and the sophisticated details make it difficult to say that they 'don't' have the problems that existing systems have, so it is seen as an argument that the problems no longer exist - not because they are gone forever, but because they just became so difficult to discuss because so much has changed ). But I think it is a problem worth looking at.
* painting the bicycle shed would be its translation into Spanish. More info here y here
And with that he momentarily extinguishes the flame (because Linus also made his flame together with Tanembaun regarding monolithic cores and micronuclei).
Linus always enjoys wars 😀
I think it is more than an anger, it is a way of going out on the show: laughs
however, it is quite clear and argued what it raises.
I think Linus was lukewarm in the interview, then I see him aggressively insult Kay on the mailing lists with good reason, so I think he said all this so as not to keep increasing the flame.
It seems to me that it has enough with the kernel, the modifications requested by Google, systemd to get into things that do not correspond to it, in fact it never got into startup systems, but, I think that it should by its means, take action on the matter , it says about the log in text…. and Linus is not a person who says «well, that's right, what are we going to do, let's keep sucking», so either this gagá and he should take a step to the side, he doesn't even write code or he's going to do something, everyone modes ……. In servers 1 cucumber matters to me what he creates, if ALL the distro are passed to systemd, it will be necessary to go to BSD, that simple, I do not care what he does or not, just as I care little what happens with Hurd, that by the way is nothing.
Anyway, buy a lot of petroleum jelly, because the time will come when RPM is a dependency on systemd, just like now they want to put brtfs ... my god ... I don't know if Google, Microsoft or redhat is worse
SynFlag = "bikeshed painting"
There is still Gentoo and Slackware, it looks like they will continue to use their default init systems for a long time.
Hopefully. I think it is time to standardize.
I do not know what the hatred is with the RPM, when as I understand Stallman himself preferred them back there. I don't care if DEB, if RPM, yes ... whatever, but a standardization is good for everyone. As Linus said, this is the time for Linux Desktop users, who care little about the technical differences between DEP and RPM as long as you double-click it and it installs.
You were going well, but you screwed up in the end machine
About RPM and SystemD… let me remind you that it exists Alien in order to convert the .rpm packages into the package you want.
Now regarding BTRFS, it is just a suggestion, because there are people who are really reluctant to use other file systems that are so mediocre that it reminds them of FAT32 or NTFS, and they prefer more robust systems (like XFS) or those suggested by majority (EXT4).
: ') as I bench Torvalds 😀, as he rightly says, and I fully share, several of the arguments against Systemd are philosophical, and as I said before, now “everyone puts on the shirt of BSD, Solaris, Haiku, HURD and how much weirdo is out there "... following the" philosophy "of Torvalds who does not like, fork, or create something new ... ..
Some will screw me, and not only am I a Linux user and RedHat does not pay me a handle xD, but I leave this to light a little more the flame http://www.phoronix.com/scan.php?page=news_item&px=MTI4NDc
and I especially love this phrase, dedicated to those who only think that Systemd is to gain speed in Boot Time «Systemd Is About Speed - Lennart says that systemd's speed is just a side-effect of designing things right…»
and sumo, some of the inits, or sets of applications (I say because since it "does not correspond" to the init) allow this:
http://diegocg.blogspot.com/2014/02/por-que-kdbus.html
http://diegocg.blogspot.com/2014/07/avanzando-golpe-de-actualizaciones-de.html
and regarding the inclusion of btrfs .. they complain but did they see why? …. http://www.phoronix.com/scan.php?page=news_item&px=MTc3NzU
Errata: «and I am not only a Linux user and RedHat does not pay me a handle xD»> «and I am only a Linux user and RedHat does not pay me a handle xD»
btrfs I really like how it works, its functions, but it is not as good as ext4 to repair itself when it gets corrupted, I had to reformat it, I was able to recover the information, but not repair the system. On the other hand, the same thing happened to me with ext4 and with fsck ready, and like new.
The problem is that nobody wants to fork SystemD like Theo de Raadt did with OpenSSL. The truth is that unfortunately no one understands reasons for not making a minimum effort to improve SystemD as a collaborator of this project.
Anyway, now I will update Opera Blink to the beta branch.
And voila: upgraded to Opera Blink 25 beta.
@amulet_linux:
Besides, on my netbook I have XFS and so far, it consumes less battery power on Debian Wheezy. The truth is that BTRFS is still green in many respects, so I opted for a more conservative solution for laptops like netbooks.
Anyway, Ext4 is good, but I can't cope with portability.
And the full glossary ideal for a good flame.
And why would Red Hat pay you? you are fucking me
There is always someone who does not understand what they want to imply
HAHAHA .. go with two balls: The Kernel is the best and SystemD a shit project .. This guy (Torvalds) if he knows how to use the words hahaha.
In Christian: he is a Pro-Unix although he likes graphical interfaces.
actually more than pro unix is a friend of what works. It will take from Unix what it works and if the rest does not serve to face current problems or needs and you have to use something radically different, it does not bother you in the least
A billion times this. All those who always accuse some software of violating UNIX principles, I hope they are doing it from a bash script that downloads web pages with wget, reads them with less and posts through curl, otherwise their opinion has no validity.
CC-Combo Breaker!
This is more of a fatality XD. The UNIX philosophy is fine for some things, but I think that, since it is speaking for a web browser, everyone follows it, they are not something inseparable monolithic, they are an amalgam of http client, html engines, javascript interpreters, plugins, SSL … They all do a single task and they do it (more or less) well.
Yes, but did you know that browsers like firefox save their download history in sqlite? and sqlite is a… BINARY FORMAT! SOMEONE CALL DENNIS RITCHIE, FIREFOX IS GOING TO EAT MY DOG!
FUD FUD FUD, the only thing that talks about SystemD.
@juanfgs
I begin by clarifying that I agree that it is time to replace init, and that systemd seems to me the best we have today for it.
But that is not why we are going to deny its deficiencies, to accuse anyone who exposes them of generating FUD, much less to fight FUD against FUD.
SQLite is a database, the format that this data contains is up to the developer, you can have int, text, which perfectly go as plain text, or BLOBs that are binary.
But for the curious case that you mention, you are somewhat lost, you can easily open the Firefox file with a text editor and see the addresses as plain text, that is, all piled up because the editors do not know the structure of SQLite, but yes You want to read them ordered in columns, from Firefox you download a plugin called SQLite manager and you see them without problems.
Unlike what happens with SystemD that forces you to use journald to be able to read it.
You have to accept sales and problems, so you can work to solve the latter.
This is false, I invite you to try it, some other string will appear on the screen, obviously but SQLite is a binary format like thousands of other binary formats that we use every day. It is unreadable to the naked eye. I don't know where you get that you are going to see addresses as plain text, the strings will jump but you will not be able to read the file in plain text mode in a reliable way.
By your own logic, SQLite is FORCING you to use some software to read its content, and that is because binary files are like that. That only journald exists to read binary logs from SystemD simply means that there is only one reader implementation of that file format, not that Lennart, Torvalds, and the Illuminati are trying to convert linux to windows as the Boycott to SystemD campaign suggests.
The problem is when we think that the binary file is "the devil", what I am asking you in case you do not notice is that if we apply the same philosophy of the SystemD detractors to other systems it falls apart. The reality that projects like SystemD or others have to be analyzed from other software perspectives, not from that of "it breaks with what we have always been doing, and therefore it is the right way." Thanks to that mentality we spent a long time without even rethinking the possibility of replacing init (then the people of runit and many others began to rethink this).
@juanfgs
I don't know if you are familiar with the concept of a database, but it is something that does not have a format, they have many TYPES OF DATA, and these are the ones that have.
You can perfectly have text strings, digits or even images in the same database. Each one with different formats.
With that in mind (I hope) you understand that it is a mistake to say that SQLite is a binary format, since it is not a file as such, you are not talking about a word document, a .pdf or a .jpg.
Now, I find it funny that you invite me to try something that you can see that you have not done.
Here are my evidences:
http://i.imgur.com/zR7PEWl.png
Yours?
It is not that "some other string" is seen, they are the complete addresses and they are readable. If it is clear to us what a database is, we will know that it is correct to say that in this case "THE DATA IS STORED AS FLAT TEXT".
I don't know how reliable you want that.
That simplistic logic that you attribute to me is because you misinterpreted my words.
If I followed it I would say that even plain texts FORCES you to use some software to read their content, and it would be true, or how would you read them without using Kate, Notepad, nano ...?
What you didn't understand is that SystemD logs can be read ONLY WITH journalct / journald. Unlike the history of Firefox that you can read it messy with any text editor or neat with a wide variety of programs:
http://www.sqlite.org/cvstrac/wiki?p=ManagementTools
I repeat in case you did not read the first time:
"I agree that it is time to replace init, and that SystemD seems like the best we have today for it."
And it is that I fully understand the advantages of using binaries (Less risk of data corruption, faster searches ...) but also its disadvantages (Dependencies, much more difficult to recover in case of data corruption ...) Denying those disadvantages is, if not but at least as absurd as not considering the advantages.
That you can see the strings does not mean that the file is parseable. Don't let Kate show you the characters in between, either.
http://imgur.com/GfUxpcf
In either case the specifications for systemd journal files are public, so writing your implementation of a journald reader is trivial, even if you don't want to use the C API they provide for that case. As well as for countless open source binary formats, APIs and libraries are provided for processing in your program.
Advantages and disadvantages, yes of course they have them, but that's where the subjective comes in, I find it obscene how the detractors jump to exaggerate it to "systemd will absorb everything", "they are FORCING us to use systemd", "linux will be windows", "bill gates will enter your house and kill your family ”, and that is FUD, you put it as you put it, it does not attack the problem in question but tries to put fear.
@juanfgs:
Are you fucking kidding me?
I see well the SQL statements in the binary in the .sqlite file in GNU Nano. And quoting @staff:
This happens due to not knowing how to handle MySQL / MariaDB.
PS: MS SQL Server is worse.
I know how to read / write SQL, basically it is a pain in the ass to have to read a sqlite in text mode for any practical use that is what I mean, and it is worth as much in that case as a binary file, you need tools to access it in a practical way and the tools They're available.
The same for systemd logs, the binary specification is available and you can write a reader (even though it is not recommended) as you see fit, or use the existing tool to access it, or use the existing C API.
And if we want to put examples of other binary files you have the rpmdbs and a lot of other applications that save settings, caches and others in binary format (MariaDB also does it optionally for the advantages that @staff mentioned), but this discussion it became a talk about avoiding the point of the conversation in the most technical way possible.
@juanfgs
“That you can see the strings does not mean that the file is parseable. Don't let Kate show you the characters in between, either. »
That is another matter completely, you said it was binary format and it has already been shown, even with your own capture, that this is false.
Now you change your argument from "It is binary and what it says cannot be read" to "It is binary it can be read but it cannot be parsed", and again you fall into an error, since the reason why you cannot analyze systematically is: why you do not know its structure.
For example:
Take a CSV file in perfect plain text, open it with Kate and you will see that it is not ordered in a readable way either. For this you need to know its structure.
So I think the problem here is that you are not very clear about what databases are, or the difference between a format and an encoding.
«… So writing your implementation of a journald reader is trivial…»
Nothing trivial, to begin with because it is not about writing a reader. If you don't have a dependency, you can create a thousand readers but journalctl / journald are MANDATORY, because they are the ones who write those journals.
"And that's FUD, whatever you put it, it doesn't attack the problem at hand but tries to scare you."
I made it very clear from my first comment that there is a lot of truth, there is FUD against SystemD, but what you do not see is that you are doing exactly the same thing.
You are generating FUD, saying things that are not true on separate topics, instead of attacking the technical question.
To say that SystemD has X problem but it is OK because many other components of the system also have it, is a terrible fallacy.
«Evil of many, consolation of fools», they say around here.
And it is that it seems incredible to me that as a proSystemD that I am, I have to be defending the "detractors who exaggerate" from the "promoters who minimize."
* Semantically, not systematically.
An apology.
I agree, how they break the balls with those cheap philosophies of 20 years ago. Sure in his time someone said, but since they are going to go from binary to a written language, the compatibility is not going to be 100%, and we are still here. Follow the progress and if you don't like them creating a fork or sending their ideas to Fedora, let me know.
At least Red Hat is a corporation that knows how to listen (not like Microsoft or Oracle), and Theo de Raadt has more balls than any son of his mother who can think of being against a project that he is not even interested in contributing.
Notable examples of notable forks: MATE (born from the defunct GNOME 2), Cinnamon (arisen from the incompatibility of users with the GNOME 3 UI paradigm), LibreSSL / BoringSSL (born thanks to the OpenSSL bug, similar to the maintenance status SystemD and its current situation), LibreOffice (emerged from the lack of OpenOffice maintenance by Oracle after the acquisition of Sun Microsystems), MariaDB (similar situation to LibreOffice, but with MySQL), Audacious (born to from the demise of XMMS), and so I can go on naming good examples of forks.
1. What are those UNIX Principles that everyone talks about (and seems to force you to use bash scripts, etc)?
2. And in any case, they cannot blame anyone for having or following those principles except themselves, because nobody sent Linux to smell the worst of UNIX or something, nor did anyone send them not to invent their own paradigms but to be copying and cloning those of another. Now are you going to complain and be ashamed of that? If that's the case, that's fine with me, but they'd have to wipe out and start over on another boat, because Linux has always been a 'wannabe Unix', that's how it was born, that's how it grew and that's how it lived.
I think they refer to:
- Do only one thing, but do it well.
- Everything in a text file.
search the wikipedia kiss principle:
“The KISS principle states that most systems work better if they are kept simple than if they are made complex; therefore, simplicity must be kept as a key design objective, and any unnecessary complexity must be avoided. "
Thanks for the response of both, now I will know what you mean.
And speaking of SystemD, I have come across this function table used for those who take the word of those who want to "boycott" SystemD (also Fedora has yours and in Spanish).
and with this it is shown that systemd is here to stay and that in fact linus supports it in a certain sense thanks to standardization since it gives real and effective solutions to real problems. it is based on practicality, not cheap philosophies from 40 years ago !!!! (philosophies that are irrelevant to current scenarios)
Standardization is fine, but if you find a comment in the systemd code, go quick to make a wish. It's not serious, I know of a project called OpenSSL (I don't know if it sounds familiar to you) that was in a similar situation, and you know what happened.
nothing to see. openSSL was maintained by a very small team of volunteer developers in their spare time and until the linux foundation reached out, it had little budget. systemd, on the other hand, has the backing of hundreds of independent developers, solid financial support from several companies that put paid developers (and it is not only han network, intel, samsung and Google have developers there as well) so they are 100% different cases. the systemd code is much more audited and has much more support.
Now the pronouncement of richard stalman and the fsf on systemd are missing which could end the discussion or blow everything up 😀
I still think that bringing up the subject that systemd is only for linux the vast majority of the objections are mere questions of a political nature, which are like a breeding ground for a certain group of fanatics who love conspiracy theories and all that bullshit.
I also think that if systemd instead of having been developed by the controversial Lennart Poettering, people from the fsf circle had made it, nobody would argue with systemd.
Looking at the comments of phoronix in the link that passed
http://www.phoronix.com/forums/showthread.php?105607-Lennart-Poettering-Talks-Up-His-New-Linux-Vision-That-Involves-Btrfs
I must confess that my belly hurt from laughing so much ... I found this comment:
08-31-2014, 10:17 PM # 9
atari314
atari314 is offline Junior Member
Metastasis… the development of secondary malignant growths at a distance from a primary site of cancer…
There are people who have a gift for saying things in a very direct and analogous way .. 🙂
And Linus, I think he does not want to create controversy and that in the end they see him as the one who sends the hell to the one who is put in front of him, that is, he is taking care of what he says, because the last times many spoke pests about his ways ...
Deep down, he knows how far he is going to allow, that is why he mentions things that he does not like or that he sees badly faced.
Well people, what I was saying happened… it was obvious, the frankestein was becoming very dangerous and bulky… so FORK !!
If gentlemen, it is called uselessd, you can interpret it as useless or as "using less", properly named.
Uselessd: A Stripped Down Version Of Systemd
http://www.phoronix.com/scan.php?page=news_item&px=MTc5MzA
Read the questions and answers on the project page, they will even make it compatible with freebsd.
http://uselessd.darknedgy.net
Now I just hope that more people join and among all the non-paid programmers by companies they achieve the long awaited replacement of init.
If, as an alternative, Devuan came out and there are certain distros that do not implement it, something will have systemd that a certain part of the community does not like.
And I don't think it's because I'm a purist.