|
|
Subscribe / Log in / New account

Wayland and Weston 1.6.0 released

Wayland and Weston 1.6.0 released

Posted Sep 19, 2014 17:11 UTC (Fri) by xtifr (guest, #143)
Parent article: Wayland and Weston 1.6.0 released

"Yes, we broke [the protocol] again since 1.5.0." And yet, they only bumped the minor version number? Tch. If the protocol is still changing that rapidly, perhaps they should still be using 0.xxx version numbers. Otherwise, breaking changes should mandate a change in the major number. Or does this protocol not count as part of the API?

They certainly can't claim ignorance of what the parts of a version number should represent. These guys have been around the block. So, what the heck is going on?


to post comments

Wayland and Weston 1.6.0 released

Posted Sep 19, 2014 17:14 UTC (Fri) by ovitters (guest, #27950) [Link]

They're talking about the xdg-shell extension, not wayland protocol.

Wayland and Weston 1.6.0 released

Posted Sep 19, 2014 22:00 UTC (Fri) by micka (subscriber, #38720) [Link] (33 responses)

> They certainly can't claim ignorance of what the parts of a version number should represent.

They are the ones that do the numbering. They get to decide the meaning of the number and its part.
A version number is purely arbitrary. Most people agree they should be increasing, that's all.
Some people assume a semantic on version number and think everybody should follow theirs.

As I understand, the only important part (wayland) has no breaking change, so does not need to increase "major" in an imaginary semantic versioning scheme. And the example implementation of a compositor (the thing that is like the twm for window managers) follows wayland version.

Wayland and Weston 1.6.0 released

Posted Sep 19, 2014 23:55 UTC (Fri) by HelloWorld (guest, #56129) [Link] (32 responses)

The semantics that most people expect from version numbers are those documented on semver.org. The fact that many projects don't follow those semantics doesn't affect that. And calling weston “like the twm” just displays blatant ignorance.

Wayland and Weston 1.6.0 released

Posted Sep 20, 2014 0:28 UTC (Sat) by raven667 (subscriber, #5198) [Link] (23 responses)

semver.org is a great standard but I'm not that confident people have any strong expectation that arbitrary projects follow that scheme, although the Wayland protocol does follow it AFAIK, they haven't made protocol-breaking changes since 1.0. The kernel definitely doesn't, they bumped to 3.x just for fun, the actual binary compatibility for userspace goes back to at least 2.4.x if not 2.0.x.

I'm also not sure why the twm comparison wouldn't be spot-on, Weston is a fully-functional demo compositor that comes with the Wayland, much like twm is a fully-functional demo window manager that comes with X. In both cases you probably use something else but it's there if you want it and makes a reference of how it's supposed to work.

Wayland and Weston 1.6.0 released

Posted Sep 20, 2014 4:14 UTC (Sat) by rahvin (guest, #16953) [Link] (22 responses)

The last person to break userspace got told off by Linus and told Linus wouldn't accept his patches until he got his act together. Last time I read the policy it was a minimum of 10 years to change the kernel userspace API in a way that breaks anything and they would only do it if there was a darn good reason.

My understanding of the commitment Wayland has made is essentially the same thing, they won't break the API. They might add features but the features already implemented will continue to work the way they do.

I agree with your Weston/twm analogy. In the long run no one is going to use Weston. Gnome and KDE and others are all writing their own Wayland compositors into their own window managers. You won't run Weston on Wayland, you'll run Gnome, KDE or some other window manager on Wayland just like you run them on X now. Weston is probably where they all get at least their initial code from but I doubt anyone will run Weston long term and that's exactly what they expect to happen as it's in the FAQ.

Wayland and Weston 1.6.0 released

Posted Sep 20, 2014 4:45 UTC (Sat) by atai (subscriber, #10977) [Link] (18 responses)

Huh? I still run twm sometimes

Wayland and Weston 1.6.0 released

Posted Sep 20, 2014 8:37 UTC (Sat) by epa (subscriber, #39769) [Link] (17 responses)

Remember when twm was too memory-hungry, slow and bloated?

Wayland and Weston 1.6.0 released

Posted Sep 20, 2014 17:13 UTC (Sat) by rsidd (subscriber, #2582) [Link] (16 responses)

Compared to what? (I only used twm on fresh installs when nothing else was available)

Wayland and Weston 1.6.0 released

Posted Sep 20, 2014 19:23 UTC (Sat) by epa (subscriber, #39769) [Link]

The original selling point of fvwm was that it used less memory than twm.

Wayland and Weston 1.6.0 released

Posted Sep 22, 2014 4:56 UTC (Mon) by eru (subscriber, #2753) [Link] (14 responses)

Compared to what? (I only used twm on fresh installs when nothing else was available)

uwm ("Unix Window Manager"), of course. It was the only WM you got out of the box in X11R3 and older. I used to run in on a Vaxstation. It did not waste screen space for title bars.

Wayland and Weston 1.6.0 released

Posted Sep 22, 2014 6:46 UTC (Mon) by neilbrown (subscriber, #359) [Link] (13 responses)

> uwm ("Unix Window Manager"),

I think you mean the "Ultrix" window manager. "Ultrix" being "DEC"s version of Unix. "DEC" being the company that made the PDP that Unix was developed on of course, and was then bought by Compaq, which was bought by HP, which was bought by... oh no, there are still around.

Wayland and Weston 1.6.0 released

Posted Sep 22, 2014 15:29 UTC (Mon) by sfeam (subscriber, #2841) [Link] (12 responses)

But he said "on a VAX". Ultrix did not run on VAX, it ran on DECstations, which had a MIPS CPU. People who ran UNIX on VAXen generally used BSD. My recollection is that by the time DEC was pushing DEC-branded UNIX (OSF/1, later Digital Unix) the window manager was DECWindows.

Wayland and Weston 1.6.0 released

Posted Sep 22, 2014 15:32 UTC (Mon) by rsidd (subscriber, #2582) [Link] (5 responses)

I note that the last three comments above come from old fogeys with tiny subscriber numbers. Mine is small too but I never used vaxen!

Wayland and Weston 1.6.0 released

Posted Sep 22, 2014 17:28 UTC (Mon) by cesarb (subscriber, #6266) [Link] (4 responses)

Low LWN.net user account numbers (they don't change whey you subscribe or stop subscribing) only mean you have used Linux for a long time. They say nothing about having used older variants of Unix.

(Linux user since 1997, which was my first real contact with Unix; and yes, I did use twm back then.)

Wayland and Weston 1.6.0 released

Posted Sep 22, 2014 19:53 UTC (Mon) by madscientist (subscriber, #16861) [Link] (3 responses)

Actually they mean you have had an LWN account for a long time. I've been using Linux since 1993 (Linux Counter # 5235).

Wayland and Weston 1.6.0 released

Posted Sep 22, 2014 22:05 UTC (Mon) by jonabbey (guest, #2736) [Link]

I used X on a Vax, and on early 90's HP workstations and SparcStations. I ported the X11R4 client libraries and utilities to the Alliant mini-super computer we had at the lab back in 1991 as well.

I used twm and liked it. ;-p

Wayland and Weston 1.6.0 released

Posted Sep 24, 2014 23:37 UTC (Wed) by nix (subscriber, #2304) [Link] (1 responses)

I'm more impressed that you can remember or have kept track of your Linux Counter number across more than twenty years than that you were using Linux in 1993, honestly :)

Wayland and Weston 1.6.0 released

Posted Sep 25, 2014 12:20 UTC (Thu) by madscientist (subscriber, #16861) [Link]

They send out an email every 5 years or so, for this or that :p :)

Plus I did have to go look it up. Every time I do I have to confess I'm mildly surprised they're still around...

Wayland and Weston 1.6.0 released

Posted Sep 22, 2014 22:21 UTC (Mon) by neilbrown (subscriber, #359) [Link] (2 responses)

The device actually mentioned was "VAXstation" which is a whole lot smaller than a "VAX", though presumably used the same instruction set.
I never actually ran Ultrix on a VAXstation (only the DECstation) but the wikipedia page for Ultrix suggests and it ran on VAX hardware.

Nonetheless, there is no "Unix Window Manager" but there certainly was an "Ultrix Window Manager": http://en.wikipedia.org/wiki/Ultrix_Window_Manager

Wayland and Weston 1.6.0 released

Posted Sep 23, 2014 0:14 UTC (Tue) by sfeam (subscriber, #2841) [Link] (1 responses)

Yeah, the VAXstations were cute. I had a "VAXstation I", sort of a prototype included as a freebie with purchase of large VAX installations. It was quite useful for driver development since it bypassed interfering with jobs on the big VAXen, but wasn't terribly useful for much else.

Either my memory is fallible or DEC marketing is incomprehensible, most likely both. I distinctly remember that the Unix I ran on the PDP-11 and early VAXen was Version 7. The Wikipedia article tells me that at some point Digital retroactively relabeled this as Ultrix. I don't think that relabeling ever penetrated my consciousness, as the first time I recall encountering an OS of that name was years later on the DECstations.

VAXstations

Posted Sep 23, 2014 14:03 UTC (Tue) by cruff (subscriber, #7201) [Link]

I have a VAXstation 4000 model 60, which is clocked at 55 MHz running VMS 7.3. The DECwindows (aka X) installed on it is not terribly speedy, partly due to the slow 8 bit frame buffer, but definitely recognizably X.

Wayland and Weston 1.6.0 released

Posted Sep 23, 2014 15:20 UTC (Tue) by jg (guest, #17537) [Link]

X and X11 certainly ran on VAXen, under Ultrix, BSD, and even VMS!

I expect that X11 apps on such a machine, if you can find one, should still work.

If someone has an old VAX around (or DECstation 3100) it would be fun to verify that this is still true.

Jim Gettys

Wayland and Weston 1.6.0 released

Posted Sep 24, 2014 6:59 UTC (Wed) by eru (subscriber, #2753) [Link]

Ultrix did not run on VAX,

It did. As I recall, it was just a kind of re-branded BSD at that time.

Wayland and Weston 1.6.0 released

Posted Sep 28, 2014 18:07 UTC (Sun) by vonbrand (subscriber, #4458) [Link]

We had Ultrix on a VAXstation and a microVAX way back then. Perhaps they weren't "classical" VAXen, but MIPS they weren't.

Wayland and Weston 1.6.0 released

Posted Sep 20, 2014 5:34 UTC (Sat) by thedevil (guest, #32913) [Link] (2 responses)

"Gnome, KDE or some other window manager"

It's my turn to call ignorance. And it's not a side issue either.

Wayland and Weston 1.6.0 released

Posted Sep 20, 2014 6:19 UTC (Sat) by raven667 (subscriber, #5198) [Link] (1 responses)

GNOME had metacity and now gnome-shell does the job, KDE has kwin, I'm not sure how making the abstract point that GNOME and KDE have their own Wayland compositors aside from Weston is "ignorance" and "not a side issue".

Wayland and Weston 1.6.0 released

Posted Sep 20, 2014 9:55 UTC (Sat) by carenas (guest, #46541) [Link]

semantics, semantics, semantics

on the other hand, I wonder if there is a registry of Wayland compositors to chose from; something like pekwm without having to run X, would be IMHO, a great user experience for people that value simplicity and nimbleness over features for a "workstation"

Wayland and Weston 1.6.0 released

Posted Sep 22, 2014 14:46 UTC (Mon) by tjc (guest, #137) [Link] (5 responses)

> The semantics that most people expect from version numbers are those documented on semver.org.

Most people! I don't know how anyone can summon the requisite chutzpah to speak on the behalf of "most people." My libertarian tendencies are sending up red flags.

In any case, most people don't even know that website exists, so it's unlikely it represents their expectations, at least not at that level of detail.

Wayland and Weston 1.6.0 released

Posted Sep 22, 2014 21:59 UTC (Mon) by xtifr (guest, #143) [Link] (4 responses)

I knew about the basic conventions *long* before that website existed--in fact, they predate the web. It's not like this is something out of the blue--there's a lot of tools and infrastructure designed to work with these conventions, and, while ignoring them for the sake of some "libertarian" ideals is certainly possible, it really just means more work for everyone, including—all-too-often—unreliable, unstable manual ad-hackery to take the place of those well-established and well-tested tools. As tends to happen in the more common case when they're ignored out of ignorance and/or stupidity.

I agree that "most people" probably aren't familiar with the conventions, but most *nix developers are, and those that aren't I would have severe concerns about hiring. It's a question of basic competence.

I do agree that semver seems a bit overspecified. But not horrifically.

That said, this particular case (Weston) seems to be justified, if, as someone else suggested, the protocol in question is already marked experimental.

Wayland and Weston 1.6.0 released

Posted Sep 23, 2014 9:09 UTC (Tue) by Wol (subscriber, #4433) [Link]

Well, a load of the software I use seems to use yyyy-mm as its version number ...

And the semantics I was originally used to, and !like! is

major-number dot minor-number bugfix-letter

I can't remember the details, but anything that shared the same major/minor was supposed to work together. Every time a new compile was pushed out for whatever reason, the bugfix got bumped. So you generally had a mishmash of bugfix letters, but you didn't mix different minors.

Cheers,
Wol

Wayland and Weston 1.6.0 released

Posted Sep 24, 2014 15:46 UTC (Wed) by tjc (guest, #137) [Link] (2 responses)

> and, while ignoring them for the sake of some "libertarian" ideals is certainly possible ...

You misread my post. The libertarian part of me is alarmed that people would presume to speak on behalf of "most people." This has nothing to do with version numbers.

Wayland and Weston 1.6.0 released

Posted Sep 25, 2014 19:51 UTC (Thu) by xtifr (guest, #143) [Link] (1 responses)

So, would you also object on libertarian grounds to a statement like "most people avoid gets()"? Or (more analogous) "most people expect a C header file to have include guards so things don't break if you include it multiple times"? That seems to me to be a simple (and accurate) observation. And so does "most people expect [these semantics] from version numbers".

I don't see anywhere where HelloWorld was saying what people *should* or *must* expect. He described his observation of what people *do* expect. And I basically agree with his observation, with the caveat that the definition on semver is a bit overspecified. (But not a lot.) I don't see how observing what people actually do violates any libertarian precepts. Which is why I was confused by your statement.

Wayland and Weston 1.6.0 released

Posted Sep 25, 2014 21:06 UTC (Thu) by mpr22 (subscriber, #60784) [Link]

I don't think most people expect any particular semantics at all from version numbers; if forced at gunpoint to state what I think most people expect from version numbers, I'd probably say something like "they go up over time" and "version 1.0 has more bugs than a bait store".

Wayland and Weston 1.6.0 released

Posted Sep 26, 2014 7:22 UTC (Fri) by mbunkus (subscriber, #87248) [Link]

> The semantics that most people expect from version numbers are those documented on semver.org.

Seems I'm not most people. Reality has led me to base my expectation on the program in question and its history of version number use. For Firefox, Chrome and similar things I don't expect API breaks with bumps in the major versions, I just expect those numbers to get higher on a roughly regular, date-based release schedule. With other programs I usually expect major new functionality if the major version number is bumped – but again not necessarily broken APIs (whatever the API constitutes for e.g. GUI programs).

And then there's Linux which, as someone already mentioned, seems to have one of the best user-space ABIs around and bumps the major version number whenever Linus thinks it's time for it.

semver.org is a nice theoretical concept for programmers, but reality doesn't care. Nor do non-technical users as far as my experience with them goes – they just expect that newer <=> higher version number. They also often hope that higher version number means fewer bugs, but we all know how true that is ;)

Wayland and Weston 1.6.0 released

Posted Sep 28, 2014 23:44 UTC (Sun) by flussence (guest, #85566) [Link]

Wrong!

An odd minor version means a development release and an even minor version is a stable release, except for when it's KDE instead of GTK where a minor of 0 indicates an alpha release and double digits means usable but EOL, or when it's Linux and a major version of 3 means the odd/even rule no longer applies, or when it's a distro where the version is commonly misinterpreted as a base-10 fraction when it's actually a YY.MM specifier, or a web browser where the version number is akin to a Unix timestamp with a granularity of six weeks and some ill-defined epoch, or the X11/OS X variant where the major version changes so infrequently it becomes part of the name itself, or glibc from the dark ages where there were no versioned releases at all, or when the length of the version number is directly proportional to bugs fixed as in TeX/Metafont, or when major versions simply disappear for marketing reasons as in Netscape and PHP, or ...

Most people (I'd like to say "all", but threads like this give me doubt) are human beings, not mindless automatons. They expect version numbers to have *useful mnemonics*.

1.6 means there's been six large updates since the project declared itself usable to a general audience. It means programs built on 1.5 likely still work. I didn't have to go memorise the meaning of those numbers from some dry spec on a distant website to understand that.

Waving a spec around is not a substitute to RTFM. Which — had it been done by the OP in the first place — would prevent this entire comment thread from existing, because as others have helpfully pointed out, that technically-correct pedantry up there isn't even factually correct. That makes it doubly worthless.

Wayland and Weston 1.6.0 released

Posted Sep 20, 2014 10:12 UTC (Sat) by tau (subscriber, #79651) [Link]

xdg-shell is not a stable protocol extension yet! you even have to explicitly invoke a "use_unstable_version" method on it with the version that your client was compiled against (which must precisely match the version that the compositor was compiled against) to make use of it.

This is also why it is being carried by downstream compositors and toolkits at the moment instead of being carried in upstream Wayland. Once it's stable then it will get upstreamed and frozen, and its out-of-band version negotiation mechanism will go away in favour of Wayland's existing built-in interface version negotiation mechanism.

xdg-shell is intended to eventually replace the core protocol's wl_shell interface which was found to have a racy design (racy in the sense that it causes display corruption, not memory corruption, thankfully). More details here:

http://blog.mecheye.net/2014/06/xdg-shell/

Protocol spec:

http://cgit.freedesktop.org/wayland/weston/tree/protocol/...


Copyright © 2025, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds