I can't believe that on the "News for nerds" site, the above comment is hiden somewhere with score 0. Yes, LWN is about the only subscription which is not only worth paying for, but where you can literally see that the money is spent on a good thing, on an effort of tracking the changes, documenting things, etc.
I guess that amongst 72M iPhones sold in 2011 you can find at least 17M identical ones. Apple sells only few models at a given time, and the variability is not that high (the SoC sourced from two different foundries is probably the biggest difference).
I highly doubt C64 is the best-selling computer of all time. Wikipedia estimates 10M-17M C64s were sold. It of course depends on what is a computer: for example, many smartphones have CPU(s), memory, storage, and even display. According to this page, in 2011 Apple sold 72M iPhones: https://www.statista.com/stati... . Also, 10M Raspberry Pi computers were sold till 2016: https://www.raspberrypi.org/bl.... I guess Arduinos have similar numbers, but they are hard to track because of clones.
The magnetic strip can easily be erased by a strong magnet (e.g. a neodymium one from a broken HDD). I erased the one on my credit card myself two years ago. However, I have since discovered that there are still payment terminals in Europe, which use solely the magnetic strip. For example, the highway toll gates in Italy and France.
I was not talking about not being able to reach the other device on the third layer (IP). My point was that even though we have perfectly good _application_-layer protocols for file sharing (CIFS, which GP thinks should be blocked), we are still doomed to share data between our devices using a third-party public cloud over HTTP[s].
Why would we? There are plenty of usable protocols for service discovery, file sharing, instant messaging, etc., but because of NATs and firewalls, everybody is doomed to use HTTP[s] to some public cloud service instead. The fact that I cannot easily copy photos between my laptop and a cell phone of my friend laying on the same desk and connected to the same WLAN without coming through the remote cloud service is pretty disappointing.
I know our servers won't accept it either since they don't even listen on it, are you saying Google is unusual in not accepting IPv6 only email? 'cause I reckon that's "standard".
Yes, Google is unusual - they do listen on IPv6 SMTP, but they reject the incoming mail as possible spam way more often than when it is being sent to them over IPv4. I had the same problem, and I had to explicitly force IPv4 for outgoing SMTP to Google in my Postfix configuration.
Also, unrelated, but I feel like the GNOME 3 hate is really blown out of proportion. Sure, some users were driven away, but the exact same thing happened with GNOME 2 and people called it trash and crap and whatever else.
And they were right.
I have been using GNOME since GNOME 1 times, and I think for former GNOME users the GNOME 3 fiasco is not something unexpected, it is a logical outcome of the overall trend in GNOME development.
I remember Sawmill/Sawfish being replaced by Metacity, which even in the latest GNOME 2 releases was not able to do things which were supported in Sawfish since day 1 and still are.
I remember Galeon being pushed out of GNOME and replaced by Epiphany (seriously, did anybody used Epiphany?), and again, Galeon was more capable than Firefox (and of course than Epiphany, but no surprise here), until it bit-rotted enough to be removed from Fedora about year and half ago.
I remember GDM being rewritten for GNOME 2.20, omitting XDMCP support altogether (a display manager without XDMCP, would you believe that?) and removing the config file, in which the user previously could set his own X server options, allowing, for example, correct multi-seat support. Those features were promised to be added later, but they never were, with the notable exception of the XDMCP support. And guess what? GDM in GNOME 3 is said to support multi-seat, but it generates its own hard-coded xorg.conf for secondary seats somewhere under/run, and again there is no way to configure the xorg.conf for secondary seats.
So no, GNOME 3 has not been a surprise, at least for me. GNOME 3 has been a logical outcome of the general trend, which has been visible in the GNOME development for several years. That said, GNOME 2 was bearable for me for general use (with Galeon, xdm, and Sawfish). When GNOME 3 was released, I have finally switched to XFCE.
Sadly, this is not true anymore. Many apps today depend on things like D-Bus or PulseAudio, which cannot be easily forwarded through the X protocol connection. Add a "run only a single instance of $app no matter what" mentality to the mix, and you are screwed: the $app started on a remote machine detects that another instance (on a completely different machine but the same display) is running, and tries to forward its own command line arguments to the previously running instance. But arguments like filenames are depended on the X client machine. Oops.
Same here. After almost 10 years with GNOME, I have migrated to XFCE for F15.
For former GNOME-2 user, XFCE provides almost the same experience: it is based on GTK, and their Terminal even is based on the same widget as gnome-terminal. I have kept my window-manager (sawfish), so the user interface is almost the same.
Unfortunately, in F15 the Galeon browser is no longer provided, so I had to migrate to Firefox, which is my biggest change in F15 from the UI standpoint. I will miss Galeon's smart bookmarks.
XFCE even supports the "desktop icons are minimized applications" mode (the default option is "desktop icons are application launchers") as the old window managers like twm or fvwm had.
no one in their right mind would ever use a CPU emulator on a mobile platform OS. It's one of the best ways to completely nuke your battery life, not to mention performance.
That's only partly true. In '94 BSD was even or ahead of Linux in terms of features.
Wrong. As one of the grandparent posts mentions, at that time Linux had shared libraries and compressed kernel, making it much more usable on desktop PCs. I know what I am talking about, I was a 386BSD and NetBSD user then, and for me those features (together with better attitude of the Linux community) were precisely the reasons for migrating to Linux.
Of course, networking at that time was more mature in *BSD. But as for overall system stability, Linux won: around 1994 it was the first UN*X I had that survived running crashme(1) on it.
I think it is not because of an incremental approach of GNOME, but rather because of their decremental approach.
Things like replacing GDM with a rewrite that still does not match the original GDM feature-wise (it even could not do XDMCP for a long time and it cannot do auto-login for single-user systems even now), replacing Sawfish with Metacity, replacing Galeon with Epiphany, which - even with epiphany-extensions package - still cannot match Galeon (despite the fact the development of Galeon has been dormant for several years now), etc.
I guess the next decremental step would be kicking out Ekiga in favour of Empathy.
How good are LVM snapshots at rolling back a hosed OS?
About the same as VM snapshots. For example, Solaris uses ZFS snapshots for OS upgrades by default.
How easy is it to:
Use the same hardware to run a test environment for that case without any virtualization products like OpenVZ/Virtuozzo or Xen/VMware etc?
It depends whether your application has any hardcoded paths in it, whether it can use e.g. different TCP ports, whether it is easy to impose ulimits on it, etc. Some can do it, some cannot. But still, it is much faster to have (say) two VMs - one for testing and one for production use, but definitely not one per application.
Just a minor nitpick: for rolling back a SW upgrade, you can use LVM/FS/storage snapshots, which are as good as the virtual machine snapshots.
And of course, many HW problems affect the whole system, the virtual machines have no means of magically escaping from this.
Don't get me wrong, I also use virtual machines for many purposes. I have just wanted to point out that a "one application = one virtual machine" approach is quite insane. We already have a means of isolating applications from each other - it is called "an operating system kernel". Virtualization just brings an unnecessary overhead in this case. OTOH, running a virtual machine with _many_ applications for the purpose of live migration and whatnot - this a right use of virtualization.
What do you do when you have to go down to the physical machine to patch firmware/bios? You lose all your applications, right?
Well, I patch the firmware probably once or at most two or three times in the server HW lifetime. OTOH, patching the OS kernel is way more frequent in my part of timespace. And the new kernel means server reboot, be it virtualized or not.
The point of virtualization is mostly to isolate the applications which require different operating systems or OS versions (with a minor added bonus of faster reboot time and live migration). But a separate virtual host per application is simply insane. After all, it is the operating system kernel which has been designed to provide a more-or-less "virtualized" view of the hardware for the applications. One OS image can more often than not run multiple applications without a problem.
There is a big problem I have with X.org server since 1.5 series (1.3 in Fedora 8 is OK): booting the secondary card. I have a dual-seat desktop (two monitors, two keyboards, and two mice), and it is not possible to use it with X.org 1.5 server, because the code for booting the secondary card via the "int10" interface has been broken since 1.5 (see bug#18160). And it seems the X.org developers have no intention of fixing it, and they instead do marginal but new things like MPX. After all, new is cool, while fixing the bugs in the old code is not.
Slashdot Mirror
This.
I can't believe that on the "News for nerds" site, the above comment is hiden somewhere with score 0. Yes, LWN is about the only subscription which is not only worth paying for, but where you can literally see that the money is spent on a good thing, on an effort of tracking the changes, documenting things, etc.
In a related news, Alphabet wants to protect its data as much as possible:
https://yro.slashdot.org/story...
It is quite interesting to see these two stories in the front page near each other.
I agree with your post. Just a minor nitpick:
> Massive duplication of data. Joining dozens of tables to get commonly needed data
You are contradicting yourself, aren't you? Keeping the schema in the normal forms means adding more joins to the queries.
I guess that amongst 72M iPhones sold in 2011 you can find at least 17M identical ones. Apple sells only few models at a given time, and the variability is not that high (the SoC sourced from two different foundries is probably the biggest difference).
I highly doubt C64 is the best-selling computer of all time. Wikipedia estimates 10M-17M C64s were sold. It of course depends on what is a computer: for example, many smartphones have CPU(s), memory, storage, and even display. According to this page, in 2011 Apple sold 72M iPhones: https://www.statista.com/stati... . Also, 10M Raspberry Pi computers were sold till 2016: https://www.raspberrypi.org/bl.... I guess Arduinos have similar numbers, but they are hard to track because of clones.
The magnetic strip can easily be erased by a strong magnet (e.g. a neodymium one from a broken HDD). I erased the one on my credit card myself two years ago. However, I have since discovered that there are still payment terminals in Europe, which use solely the magnetic strip. For example, the highway toll gates in Italy and France.
I was not talking about not being able to reach the other device on the third layer (IP). My point was that even though we have perfectly good _application_-layer protocols for file sharing (CIFS, which GP thinks should be blocked), we are still doomed to share data between our devices using a third-party public cloud over HTTP[s].
Why would we? There are plenty of usable protocols for service discovery, file sharing, instant messaging, etc., but because of NATs and firewalls, everybody is doomed to use HTTP[s] to some public cloud service instead. The fact that I cannot easily copy photos between my laptop and a cell phone of my friend laying on the same desk and connected to the same WLAN without coming through the remote cloud service is pretty disappointing.
I know our servers won't accept it either since they don't even listen on it, are you saying Google is unusual in not accepting IPv6 only email? 'cause I reckon that's "standard".
Yes, Google is unusual - they do listen on IPv6 SMTP, but they reject the incoming mail as possible spam way more often than when it is being sent to them over IPv4. I had the same problem, and I had to explicitly force IPv4 for outgoing SMTP to Google in my Postfix configuration.
> it's an acronym for 'dynamic host configuration protocol,' and is a key building block of network management.
The above explanation is a clear proof that Slashdot is not a "news for nerds" site anymore.
Probably by writing a reply.
Maybe this one will be next:
http://xkcd.com/695/
Also, unrelated, but I feel like the GNOME 3 hate is really blown out of proportion. Sure, some users were driven away, but the exact same thing happened with GNOME 2 and people called it trash and crap and whatever else.
And they were right.
I have been using GNOME since GNOME 1 times, and I think for former GNOME users the GNOME 3 fiasco is not something unexpected, it is a logical outcome of the overall trend in GNOME development.
I remember Sawmill/Sawfish being replaced by Metacity, which even in the latest GNOME 2 releases was not able to do things which were supported in Sawfish since day 1 and still are.
I remember Galeon being pushed out of GNOME and replaced by Epiphany (seriously, did anybody used Epiphany?), and again, Galeon was more capable than Firefox (and of course than Epiphany, but no surprise here), until it bit-rotted enough to be removed from Fedora about year and half ago.
I remember GDM being rewritten for GNOME 2.20, omitting XDMCP support altogether (a display manager without XDMCP, would you believe that?) and removing the config file, in which the user previously could set his own X server options, allowing, for example, correct multi-seat support. Those features were promised to be added later, but they never were, with the notable exception of the XDMCP support. And guess what? GDM in GNOME 3 is said to support multi-seat, but it generates its own hard-coded xorg.conf for secondary seats somewhere under /run, and again there is no way to configure the xorg.conf for secondary seats.
So no, GNOME 3 has not been a surprise, at least for me. GNOME 3 has been a logical outcome of the general trend, which has been visible in the GNOME development for several years. That said, GNOME 2 was bearable for me for general use (with Galeon, xdm, and Sawfish). When GNOME 3 was released, I have finally switched to XFCE.
Today all of our apps are network transparent.
Sadly, this is not true anymore. Many apps today depend on things like D-Bus or PulseAudio, which cannot be easily forwarded through the X protocol connection. Add a "run only a single instance of $app no matter what" mentality to the mix, and you are screwed: the $app started on a remote machine detects that another instance (on a completely different machine but the same display) is running, and tries to forward its own command line arguments to the previously running instance. But arguments like filenames are depended on the X client machine. Oops.
... Slashdot leaves CmdrTaco.
Same here. After almost 10 years with GNOME, I have migrated to XFCE for F15.
For former GNOME-2 user, XFCE provides almost the same experience: it is based on GTK, and their Terminal even is based on the same widget as gnome-terminal. I have kept my window-manager (sawfish), so the user interface is almost the same.
Unfortunately, in F15 the Galeon browser is no longer provided, so I had to migrate to Firefox, which is my biggest change in F15 from the UI standpoint. I will miss Galeon's smart bookmarks.
XFCE even supports the "desktop icons are minimized applications" mode (the default option is "desktop icons are application launchers") as the old window managers like twm or fvwm had.
no one in their right mind would ever use a CPU emulator on a mobile platform OS. It's one of the best ways to completely nuke your battery life, not to mention performance.
Like, er, Transmeta?
That's only partly true. In '94 BSD was even or ahead of Linux in terms of features.
Wrong. As one of the grandparent posts mentions, at that time Linux had shared libraries and compressed kernel, making it much more usable on desktop PCs. I know what I am talking about, I was a 386BSD and NetBSD user then, and for me those features (together with better attitude of the Linux community) were precisely the reasons for migrating to Linux.
Of course, networking at that time was more mature in *BSD. But as for overall system stability, Linux won: around 1994 it was the first UN*X I had that survived running crashme(1) on it.
Theoretically? I have been running a dual-seat configuration on my home workstation for about 5 years now.
I stand corrected on the autologin issue, thanks.
Another thing the new GDM cannot do is to specify the arguments of the X server command line (e.g. for running two X servers in a multiseat setup).
I think it is not because of an incremental approach of GNOME, but rather because of their decremental approach.
Things like replacing GDM with a rewrite that still does not match the original GDM feature-wise (it even could not do XDMCP for a long time and it cannot do auto-login for single-user systems even now), replacing Sawfish with Metacity, replacing Galeon with Epiphany, which - even with epiphany-extensions package - still cannot match Galeon (despite the fact the development of Galeon has been dormant for several years now), etc.
I guess the next decremental step would be kicking out Ekiga in favour of Empathy.
How good are LVM snapshots at rolling back a hosed OS?
About the same as VM snapshots. For example, Solaris uses ZFS snapshots for OS upgrades by default.
How easy is it to:
Use the same hardware to run a test environment for that case without any virtualization products like OpenVZ/Virtuozzo or Xen/VMware etc?
It depends whether your application has any hardcoded paths in it, whether it can use e.g. different TCP ports, whether it is easy to impose ulimits on it, etc. Some can do it, some cannot. But still, it is much faster to have (say) two VMs - one for testing and one for production use, but definitely not one per application.
Just a minor nitpick: for rolling back a SW upgrade, you can use LVM/FS/storage snapshots, which are as good as the virtual machine snapshots.
And of course, many HW problems affect the whole system, the virtual machines have no means of magically escaping from this.
Don't get me wrong, I also use virtual machines for many purposes. I have just wanted to point out that a "one application = one virtual machine" approach is quite insane. We already have a means of isolating applications from each other - it is called "an operating system kernel". Virtualization just brings an unnecessary overhead in this case. OTOH, running a virtual machine with _many_ applications for the purpose of live migration and whatnot - this a right use of virtualization.
What do you do when you have to go down to the physical machine to patch firmware/bios? You lose all your applications, right?
Well, I patch the firmware probably once or at most two or three times in the server HW lifetime. OTOH, patching the OS kernel is way more frequent in my part of timespace. And the new kernel means server reboot, be it virtualized or not.
The point of virtualization is mostly to isolate the applications which require different operating systems or OS versions (with a minor added bonus of faster reboot time and live migration). But a separate virtual host per application is simply insane. After all, it is the operating system kernel which has been designed to provide a more-or-less "virtualized" view of the hardware for the applications. One OS image can more often than not run multiple applications without a problem.
There is a big problem I have with X.org server since 1.5 series (1.3 in Fedora 8 is OK): booting the secondary card. I have a dual-seat desktop (two monitors, two keyboards, and two mice), and it is not possible to use it with X.org 1.5 server, because the code for booting the secondary card via the "int10" interface has been broken since 1.5 (see bug#18160). And it seems the X.org developers have no intention of fixing it, and they instead do marginal but new things like MPX. After all, new is cool, while fixing the bugs in the old code is not.
So, will Wayland work also on secondary cards?