Hacker News Reader

Comment by vilhelm_s | Hacker News Reader

And several other Mac features were stopped before they made it into the official release. This is Richard Stallman's policy: "GNU Emacs should never offer people a practical reason to use some other system instead of GNU. Therefore, when someone implements a useful new feature but only for a non-GNU system, we do not accept it that form."

https://lists.gnu.org/archive/html/emacs-devel/2015-12/msg01...

View on HN

If the goal is to encourage people to use GNU software and potentially even migrate to a full GNU system, I find this policy counterintuitive. It comes off as needless and antagonistic and it pushes me further away from GNU and towards free software that's not so tangled up in rigid philosophies, or worse, proprietary software.

> It comes off as needless and antagonistic

That's because that comes directly from the commercial companies rulebook. Apple (for example) will not make an application that runs better on Android or Windows than on one of its OS.

It would not be beside any of the big brand to remove something that was working before just because that remove a tiny bit of interest in their own platform. That makes cruel business sense ... and they are also hated for it when it happens.

If you discover Emacs on macOS, and like among other things the multicolor feature, then decide to try the Full GNU Experience™ and find that that feature is not supported on the flagship (!) GNU product, well, then maybe you’ll be back to macOS after all.

Let’s not forget that thanks to its libre licensing this is a strategic issue, but not a technical one – recovering the removed features on macOS is a few patches away.

Then, by all means, go use whatever software you want.

There's a place for ideological purists in software and I for one am glad RMS is around.

Did you expect anything tasteful or coherent from RMS?

Seems to be a lot of hate for this policy, but the goal is simply universal interoperability, which is laudible. Otherwise we might as well go back to the days of IE6, since apparently there is absolutely no problem with Embrace-Extend-Extinguish, as long as we can have emojis on our Macbooks.

Edit: I should add that, since it doesn't appear to be a true interop issue, it may not be a sound decision here. I'm defending the rationale for the policy, not this particular application of it.

> the goal is simply universal interoperability

No, the goal is that "GNU Emacs should never offer people a practical reason to use some other system instead of GNU".

Universal interoperability is just a side effect.

You've got it backwards. Stallman may be disagreeable at times, but he's no fool. It's written this way to address the root cause of interop issues, which are themselves a secondary effect.

Of course he wants people to choose GNU, but most importantly, he wants to preserve people's freedom to choose GNU, which means it must not become a second-class citizen to proprietory additions from Apple, Microsoft, or anyone else. If it's portable to GNU, the issue goes away. He says so explicitly.

And nobody will pick the GNU solution when the non-GNU solution works better regardless of the OS

Though under this wording emacs would theoretically accept a GNU OS-only feature. It just happens that GNU OSes rarely offer extra features that emacs can take advantage of.

The goal is "the lowest common denominator", while when making a cross-platform software it makes sense to have features that use the strength of each platform.

Yes, yes. Hold Emacs hostage Mr. Stallman. That'll make everyone stay up all night hacking to bring your lackluster platform to parity with its competition.

There's a real need for innovation and great ideas to better support all manner of open community resources, free software and beyond. But stupid control games are totes better than all that.

Eh.

It's RMS being RMS. I think he's ridiculously dogmatic and unpractical, but you can't be surprised by this - he has his ideologies to push with fervent absolutism. I doubt he's all that concerned about the market share of his text editor.

>That'll make everyone stay up all night hacking to bring your lackluster platform to parity with its competition.

Do you know any other platform that managed to form a foundation for org-mode? Until there is, emacs may not be called a "lackluster platform".

I'm not referring to Emacs with that, but rather to Stallman holding Emacs functionality hostage to boost GNU OS platform(s). Which are definitely lackluster.

Its not because its not shiny and not pricey that its lackluster. At least you can customize stuff on your OS instead of having to follow the only way designed by Apple.

Thank you for this. It validates and reaffirms my hate of GNU.

GNU is not UNIX, and no amount of "freedom" peddling changes the fact that it purposely violates UNIX principles. The only thing that could change that is their code (for example ditching info in favor of man pages, removing the z option in GNU tar, or dropping --long-options), and that's one thing they'll never change. I'm so glad I don't have to put up with the likes of Stallman "setting me free" by removing features GNU/Linux doesn't support.

Is there any non-dead UNIX that doesn't, nowadays? The BSD tar also has -z, as does SmartOS.

As much as I enjoy "cat -v considered harmful", it seems the battle is lost for UNIX itself.

http://harmful.cat-v.org/cat-v/

At least on FreeBSD, I'm on BSD (don't have to use GNU), and on illumos / SmartOS I'm on AT&T System V release 4.0. As long as those alternatives exist GNU holds no power over me and provides no value that I care about, or need. Both BSD and SmartOS are free software.

Remove z option in tar? Where did that come from?

From the fact that tar is a tape archiver, and according to the UNIX philosophy of "do one thing and do it well" it should empathically not dub as a (de)compressor. GNU implements tools inside of other tools. Emacs is a perfect example if that.

GNU didn't originate from a unix mindset. You should know this.

That doesn't mean their software is bad.

Also, Emacs is incredibly useful, and one of the best general-purpose text editors out there.

GNU didn't originate from a unix mindset. You should know this.

I do and oh-how-right you are. What I have major difficulty with is comprehending why people like it, and why they tolerate Stallman's nonsense when there are alternatives. The "rip it all out because it's not ours and re-invent everything just because" attitude is so fundamentale broken.

As somebody who does tolerate Stallman's nonsense, I can tell you why: It's because some of his software is really good. Like Emacs. I'm willing to bet that you're a VI user, but for us Emacs users, there really is no substitute. And if you want a Scheme package, Guile is a really good one (although it's not as efficient as some of the compiled schemes): Andy is doing a great job with it.

There are other examples, like tar (I know you would disagree, but dammit, GNU tar is so useful). When there isn't, or Stallman deliberately cripples his software, there are usually alternatives (clang, etc.) which we use, unless "we" are FSF people. Which most of us aren't.

Yes, his attitude is somewhat broken, although I understand why it exists. Also, he's not quite as NIH as you seem to think: There are several cases where he advocates use of software that isn't his own (heck, he advocates clang, or at least has no problem with it).

Also, I have no objection to long options, or to utilities that are a little bit richer (just so long as they still interoperate, and you can still use the standard options, which is usually the case).

I'm willing to bet that you're a VI user,

And you'd win that bet. Not even a vim user (except on Amiga), but straight vi. Why? Because vi was always there when I needed a text editor, and by learning vi, I automatically learned ed, sed, and ex. All of a sudden, I could write multi line sed programs.

Hey, I get it. I've always have the sneaking suspicion that VI users are way more productive at actually editing. But I know for a fact that they don't have the extensions that make editing dynamic languages (especially lisp) in Emacs so efficient. And there's no use learning another editor just to write C: What do you think I am, and IDE user?

The -z option is just a more convenient way of piping the tarball through gzip at the appropriate time. GNU tar doesn't actually contain its own (de)compressor. You can also get it to use a compression program of your choice.

In that case you mean option -x not -z.

Either way, there are several philosophies and schools of thought. Many of them yield good results in different situations. "Do one thing..." is not the same as "do _only_ one thing..." and there are exceptions to the rules and _we_ made the rules so _we_ can/may change them.

Sticking to rules when they don't make sense is stubbornness and locks one in a local maximum.

Using a dedicated (de)compressor and then piping that as input to a tape archiver makes perfect sense to me:

  tar cf - file | xz -9evc > archive.tar.xz

  xz -dvc archive.tar.xz | tar xf -

If it's a tape archiver, why is it allowed to write to disk? Check and mate.

tar isn't allowed or disallowed to write to disk -- f specifies a device, and that device just happens to be a file on a filesystem. Since everything on UNIX is a file, even a tape drive, even a huge 6,000 tape silo with 128 tape drives is multiple files under /dev/rmt/, this does not violate the principle.

To put this into perspective for you, you could use the block device /dev/dsk/c0t0d0s2 as a tar file to write a tape archive to:

tar cvf /some/path /dev/dsk/c0t0d0s2

The manual page for tar clearly states that tar is for tape archive manipulation:

  BSDTAR(1)                 BSD General Commands Manual                BSDTAR(1)

  NAME
       tar -- manipulate tape archives

Tape.

Your statement is an indication that what you really need is to read AEleen Frisch's Essential System Administration so that the UNIX concepts would sink in, because if you're making such statements, you're just not there yet.

I've been using unix since 1992, starting with ultrix, then going to sunos, netbsd, solaris, linux, tru64, osx, and I think aix was in there somewhere too.

But how much have you really learned? And were you an end user or a system administrator? I work with people who have used UNIX for 20-30 years and still don't understand why configuration packages should automatically (re)start a service upon upgrade, stop the service before removal, and start it upon initial install of the configuration package. 30 years of using all kinds of UNIX operating systems and they haven't learned a thing beyond being an end user. And some of these people are developers. So the length of usage isn't crucial, but how much one has learned about the subject -- with an understanding of concepts and ideas behind it. I used all of those you list -- but as my breadth and usage deepened, I sought to understand the principles and ideas behind UNIX.

tar is a utility for creating and extracting .tar and .tgz files.

No, gzip compressed tape archive has nothing to do with the tape archiver or the archiver format. Pipe the format to stdout without using compression and then it should be pretty obvious. This isn't Windows, Atari TOS or AmigaOS where archivers like pkzip and lha or lzx are also (de)compressors: that is precisely why on UNIX, gzip, compress, bzip2 or xz can only compress one single file and aren't archivers. They're dedicated (de)compressors. cpio, ar, pax and tar serve as dedicated archivers.

Reply on news.ycombinator.com