Oracle Switching Solaris To A Continuous Delivery Model

[ldo17]

I do not have to hexedit "fix" old Windows XP binaries for running those on Win10. Get the conceptual difference?

https://www.google.com/search?q=windows+10+incompatible

[TheBlackCat]

On Windows, it's more or less the same. You can run pretty old binaries just fine. In fact, when using 32bit Windows, you can theoretically run even ancient 16-bit applications (Win3.x)

Not if they depend on old drivers that weren't updated to work with recent versions of Windows.

[aht0]

Not if they depend on old drivers that weren't updated to work with recent versions of Windows.

I've used Windows Vista drivers for 7, 7 drivers for 8, 8 drivers for 10 and even 7 drivers on 10. Over the Windows internal compatibility layer. Probability of success is best when you stick to using drivers from most recent version of Windows possible. For some hardware I've been forced to use 7 driver (since there is nothing newer) on Win10 and have still succeeded in getting the hardware work.

It's one way for getting stuff working which did not receive "official" support for OS upgrade or was ancient to begin with. Time difference between releases of 7 RTM and 10 is far more substantial that couple of years which are already problematic in Linux.

From older ones, you can use WinXP drivers on Windows 2003 Server and vice versa. They also share their particular driver model version, which changed with the coming of Vista. Thus, what does not work is using Vista hardware drivers on WinXP/2003 and 2003/XP drivers on Vista and newer. It does not count of various x-drivers l

Clear?

[TheBlackCat]

I've used Windows Vista drivers for 7, 7 drivers for 8, 8 drivers for 10 and even 7 drivers on 10. Over the Windows internal compatibility layer.

When it works, that is great. But when it doesn't, you have the exact same problem you are complaining about with Linux. Pretty much none of my laptop's Windows 7 or 8 drivers work on Windows 10.

The same is true for, say, the audio equipment we use in our lab, despite the fact that the audio subsystems is supposed to be pretty similar since Windows Vista.

Windows and Linux are in the exact same boat here. And Linux has the advantage that most drivers are maintained as part of the operating system, so they have support for much, much longer than they do on Windows. Having a laptop from 10 years ago work out-of-the-box with the latest version of Linux is no problem (I have done it), but good luck having such a laptop work at all under Windows 10.

[aht0]

When it works, that is great. But when it doesn't, you have the exact same problem you are complaining about with Linux. Pretty much none of my laptop's Windows 7 or 8 drivers work on Windows 10.

Make and model of the laptop? Out of curiosity.

You brought out 10+ years old laptops. Ones worth having 7 or newer on them start with a Socket P. Which are almost exactly decade old or newer. On such you can generally get the 7/8/10 on, some take even 10 quite easily, some only provided you have patience. Problematic spots are the necessity to upgrade DDR2 RAM to 4Gb, waste some SSD on vastly slower early SATA interface and possibly work around driver issues with OEM branded hardware. Which is all doable.

Machines pre-dating socket P, DDR2 and SATA drives are not worth keeping nor installing 7 on it. Slow, no new 2,5" PATA supply (unless you pay like 100+euros), low amount of RAM. I can't see any sane person wanting to use 7 or 10 on such machine. It'd be like masochism. Yeah, put a XP or some linux running LXDE on it until it finally dies. Machines sporting Mobile Athlons and other such crap but having DDR2 and SATA, I've also put 7 and 8 on, AMD's GPU is the most annoying aspect but generally doable (block driver signature check and just find the closest fitting driver for chosen Win version, fucking edit the driver inf files with "correct" hw id's). After multiple BSODs one usually works it out. 8.1 drivers tend to work on 10. If there are no better than 7 and 10 does not work work with it, I'll just put 8 on it and tell owner to be satisfied it or buy him/herself a new machine.

The same is true for, say, the audio equipment we use in our lab, despite the fact that the audio subsystems is supposed to be pretty similar since Windows Vista.

I'll take what you claim as true. I've heard really professional specialist sound cards suck this way. For a hint, EMU10K1 based cards (old Creative PCI stuff) can be installed using win7 compatibility layer on 8-10.

Windows and Linux are in the exact same boat here. And Linux has the advantage that most drivers are maintained as part of the operating system, so they have support for much, much longer than they do on Windows. Having a laptop from 10 years ago work out-of-the-box with the latest version of Linux is no problem (I have done it), but good luck having such a laptop work at all under Windows 10.

[/quote]

Are they? Are you sure? You are forgetting or ignoring some essential differences. Windows drivers come always in binary form. Already compiled. How about Linux drivers? Do you get Linux "stand-alone" drivers in "1-works-for-all binary format" or you are getting the said drivers generally as sources you'd have to compile in some way before loading the driver modules?

I sorta remember all the fucking problems with Radeon Linux binary drivers which simply did not want to work even on a very distro they were supposed to be compatible with. Majority of Linux "extra-kernel" drivers I've seen are first being compiled against your particular kernel, then used. Or you are in for a lots of trouble.

Win7, 8, 8.1, 10 all have DIFFERENT kernels, kernels differ even inside the particular Windows version, marked by build versions. Win10 RTM does not have the same kernel as Win10 Anniversary Update. But still, you are able to often cross-install older drivers on newer OS. This is what backwards compatibility means.

----------------------
Linux has dumped all the backwards compatibility in exchange for fast evolution and lots of change. You can't have both at once. Not without immense effort. That's my opinion.

[TheBlackCat]

Make and model of the laptop? Out of curiosity.

Dell XPS L502x

You brought out 10+ years old laptops. Ones worth having 7 or newer on them start with a Socket P. Which are almost exactly decade old or newer. On such you can generally get the 7/8/10 on, some take even 10 quite easily, some only provided you have patience. Problematic spots are the necessity to upgrade DDR2 RAM to 4Gb, waste some SSD on vastly slower early SATA interface and possibly work around driver issues with OEM branded hardware. Which is all doable.

And completely invalidates your claim about Windows working "out-of-the-box". None of those issues you listed are a problem on Linux.

Machines pre-dating socket P, DDR2 and SATA drives are not worth keeping nor installing 7 on it. Slow, no new 2,5" PATA supply (unless you pay like 100+euros), low amount of RAM. I can't see any sane person wanting to use 7 or 10 on such machine. It'd be like masochism. Yeah, put a XP or some linux running LXDE on it until it finally dies. Machines sporting Mobile Athlons and other such crap but having DDR2 and SATA, I've also put 7 and 8 on, AMD's GPU is the most annoying aspect but generally doable (block driver signature check and just find the closest fitting driver for chosen Win version, fucking edit the driver inf files with "correct" hw id's). After multiple BSODs one usually works it out. 8.1 drivers tend to work on 10. If there are no better than 7 and 10 does not work work with it, I'll just put 8 on it and tell owner to be satisfied it or buy him/herself a new machine.

You are refuting your own claims here.

I'll take what you claim as true. I've heard really professional specialist sound cards suck this way. For a hint, EMU10K1 based cards (old Creative PCI stuff) can be installed using win7 compatibility layer on 8-10.

Uh, no, that is no where close to what we need. And it isn't a matter of the card sucking, it is a matter with Windows audio APIs sucking and Microsoft continuing to change them in subtle ways in their attempts to make them suck slightly less, which breaks everything dependent on careful timing and latency control in unpredictable ways.

Are they? Are you sure? You are forgetting or ignoring some essential differences. Windows drivers come always in binary form. Already compiled. How about Linux drivers? Do you get Linux "stand-alone" drivers in "1-works-for-all binary format" or you are getting the said drivers generally as sources you'd have to compile in some way before loading the driver modules?

I have never in my life compiled a driver from source on Linux. The only drivers I have had to compile from source were for Windows, ironically.

Win7, 8, 8.1, 10 all have DIFFERENT kernels, kernels differ even inside the particular Windows version, marked by build versions. Win10 RTM does not have the same kernel as Win10 Anniversary Update. But still, you are able to often cross-install older drivers on newer OS. This is what backwards compatibility means.

It depends on how deeply the driver integrates itself into the OS. lots of Linux drivers work just fine across versions. The difference is that, because the kernel is open-source, you can integrate much deeper into it than you can with windows. But that also has the cost that you need to keep up with the kernel. The difference isn't that Windows has stable APIs and Linux doesn't, the difference is that Windows doesn't let you do the additional stuff Linux lets you do.

[aht0]

I tried to include some related quotes from earlier posts and make it more easily readable.

TheBlackCat:When it works, that is great. But when it doesn't, you have the exact same problem you are complaining about with Linux. Pretty much none of my laptop's Windows 7 or 8 drivers work on Windows 10.

aht0:Make and model of the laptop? Out of curiosity.

Dell XPS L502x

Did you try to use Dell installer packages "as is"? If so, Dell driver utilities are locking you out. It's not exactly problem caused by MS but Dell. You can actually unpack Dell driver installers using 7zip or Universal Extractor. Both work about the same. Try importing extracted drivers manually.

Make clear distinction between issues caused by MS and intentional installer-blacklisting done by some OEM for commercial/support reasons. What drivers were you trying to import which weren't present in Win10 already?

aht0:You brought out 10+ years old laptops. Ones worth having 7 or newer on them start with a Socket P. Which are almost exactly decade old or newer. On such you can generally get the 7/8/10 on, some take even 10 quite easily, some only provided you have patience. Problematic spots are the necessity to upgrade DDR2 RAM to 4Gb, waste some SSD on vastly slower early SATA interface and possibly work around driver issues with OEM branded hardware. Which is all doable.

And completely invalidates your claim about Windows working "out-of-the-box".

What I really claimed initially (if you bothered to check) was windows binary programs generally working across different Windows versions OTB. Discussion just drifted off.
In comparison, again. TRY to run Linux binary compiled using older distribution with different shared library versions. Windows, BSD and Solaris make special effort to support it. Solaris even went to ridiculous lengths about backwards compatibility of binary code. Linux does not.care.at.all.

You brought up cross-loading Windows drivers/their lack of existence. I simply told you why you often can't find such for old hardware. It's drifting away from the main argument anyway.

None of those issues you listed are a problem on Linux

Bucket of bullshit and bunch of side-tracking. Initial discussion was over contemporary OS'es ability for running OLD(ER) binary code, not discussion over how to put contemporary OS on 10-30 years old machine. 2 utterly different things. Why you'd want to argue over that is understandable, since this area is the only one where you even have some counter-arguments working for you. But it's irrelevant, because running older programs (for example proprietary stuff) on newer OS is actually important (companies may well be unable or unwilling to change their business software, but have to eventually swap out workstations, which already come with newer Windows for example).

Uh, no, that is nowhere close to what we need. And it isn't a matter of the card sucking, it is a matter with Windows audio APIs sucking and Microsoft continuing to change them in subtle ways in their attempts to make them suck slightly less, which breaks everything dependent on careful timing and latency control in unpredictable ways.

Btw, WHY would you want to fuck around changing OS on a lab machine? Any particular reason? Besides "there is a newer OS around". I have couple of possible valid scenarios in mind but I am unwilling to provide you with free ammunition

I have never in my life compiled a driver from source on Linux. The only drivers I have had to compile from source were for Windows, ironically.

Perhaps not consciously. Isn't your Dell XPS sporting Geforce 540GT? What do you think why Linux's Nvidia's binary driver (one you can download from Nvidia's web site) requires compiler, kernel-headers and so forth before you could install it. It's pretty much source code and bunch of firmware blobs packed into self-contained shell script. Any driver outside particular kernel has to be compiled before you can load it. Installer script does it all. Or, if you are installing your drivers from package repositories, how did they end up there? Why are Linux packages dependency-tied up to ridiculous lengths?

SystemCrasher recently bitched that Nvidia's most recent binary drivers are not working with Linux 4.10. Wonder why, if there, according to you, exists such a wonderful compatibility?

It depends on how deeply the driver integrates itself into the OS. lots of Linux drivers work just fine across versions. The difference is that, because the kernel is open-source, you can integrate much deeper into it than you can with windows. But that also has the cost that you need to keep up with the kernel. The difference isn't that Windows has stable APIs and Linux doesn't, the difference is that Windows doesn't let you do the additional stuff Linux lets you do.

It depends if there have been changes in upstream, how many and which, compared to the time particular driver was released. Drivers and their related utilities break quite often because upstream devs change the kernel internals overnight.

[TheBlackCat]

Did you try to use Dell installer packages "as is"? If so, Dell driver utilities are locking you out. It's not exactly problem caused by MS but Dell. You can actually unpack Dell driver installers using 7zip or Universal Extractor. Both work about the same. Try importing extracted drivers manually.

People have tried that with the laptop and it led to much of the hardware working poorly or not at all.

What I really claimed initially (if you bothered to check) was windows binary programs generally working across different Windows versions OTB. Discussion just drifted off.
In comparison, again. TRY to run Linux binary compiled using older distribution with different shared library versions. Windows, BSD and Solaris make special effort to support it. Solaris even went to ridiculous lengths about backwards compatibility of binary code. Linux does not.care.at.all.

No, most software simply includes their own shared libraries along with the software. Windows software that uses Qt or GTK, for example, will include their own Qt or GTK release in the software. You can do that with Linux too, a lot of proprietary packages do it, and you can do it with packages you build if you want to. Those work out of the box across Linux versions at least as reliably as Windows ones. But Linux distros usually try to avoid it that because it leads to a lot of duplication, wasted space, and security issues when the bundled libraries aren't updated.

You brought up cross-loading Windows drivers/their lack of existence. I simply told you why you often can't find such for old hardware. It's drifting away from the main argument anyway.

No, we got started on drivers because your initial (and only) example so far was about a driver issue.

Bucket of bullshit and bunch of side-tracking. Initial discussion was over contemporary OS'es ability for running OLD(ER) binary code, not discussion over how to put contemporary OS on 10-30 years old machine. 2 utterly different things.

No, the discussion you started was about running drivers for an old version of an OS on a new version of an OS. My point was that this doesn't work reliably on Windows either.

Btw, WHY would you want to fuck around changing OS on a lab machine?

Now who is the one trying to sidetrack? It doesn't matter the reason, what matters is whether your example about a Linux driver not being compatible with a later version of Linux also applies to windows drivers. It does.

SystemCrasher recently bitched that Nvidia's most recent binary drivers are not working with Linux 4.10. Wonder why, if there, according to you, exists such a wonderful compatibility?

Because they dig into lower-level interfaces that aren't as consistent. Not that even Windows high-level driver interfaces are consistent.

[aht0]

People have tried that with the laptop and it led to much of the hardware working poorly or not at all.

I do it pretty often, mostly in order to put some LTE mPCIe modem in Dell machine that had originally UMTS card in it. Like yours. Dell installer would otherwise recognize that particular driver is not meant for that model and refuse installing the driver. Particularly on Win7. 8.1/10 may often recognize the broadband modems on their own. But, unpacking the driver is an option and you don't have to fuck around with USB net-sticks.

Because they dig into lower-level interfaces that aren't as consistent. Not that even Windows high-level driver interfaces are consistent.

Not "as consistent" but "aren't consistent AT ALL". It's one facet of Linux which is in constant change. Windows driver interfaces are glacially-changing in comparison. And there is at least good possibility for getting the drivers work over compatibility layer. Which is not even existing in Linux.

Now who is the one trying to sidetrack? It doesn't matter the reason, what matters is whether your example about a Linux driver not being compatible with a later version of Linux also applies to windows drivers. It does.

I was simply curious.

No, the discussion you started was about running drivers for an old version of an OS on a new version of an OS. My point was that this doesn't work reliably on Windows either.

Me?

aht0:On Windows, it's more or less the same. You can run pretty old binaries just fine. In fact, when using 32bit Windows, you can theoretically run even ancient 16-bit applications (Win3.x)

TheBlackCat:Not if they depend on old drivers that weren't updated to work with recent versions of Windows.

Let it be as you claim then. Linux is perfect in all ways..

FreeBSD, Solaris, NetBSD, OpenBSD and even Windows do bother with providing compatibility modules for earlier library versions. Linux never needs any. It always works OTB. Sarcasm.

[ldo17]

Let it be as you claim then. Linux is perfect in all ways..

Passive-aggressive, much?