FreeBSD Can Now Be Built From Linux/macOS Hosts, Transition To Git Continues
phoronix.com
phoronix.com
Their migration from Subversion to Git also provided the motive for the murder mystery book "git commit murder" by Michael W. Lucas.
In other news, my butcher has finally become a vegetarian.
But, joking aside, it's great that FreeBSD has improved Linux KPI layer, improved Linux application compatibility, and has synced their ZFS code with upstream ZFS-on-Linux.
That happened in Grenoble, France: The guy behind "La bouche'rit" ("The mouth laughs", a pun on "boucherie", the butchery), a vegan restaurant, was a previously a butcher.
Also, most conversion tools will handle the most common cases OK, and only generate garbage in oddball situations. They don't print out a warning when the oddball situations happen, so unless you go back and look at every commit you'll never notice the damage. Most of those situations only happen because of accidents, like creating a Subversion branch the wrong way, then deleting it and making it the correct way. Or adding files to a CVS repository by copying ,v files from a different CVS repository.
I'd love to be able to use a dedicated transition tool of the quality FreeBSD would like and need.
Haven't used it myself though; I just noticed that ESR has spent considerable effort in creating reposurgeon.
From [0]:
> The software used for the conversion is a slightly modified fork of svn2git, as used by the KDE project.
> It requires a rules file to map svn trees and/or revisions to git branches or tags. The current rules don't make use of tags but simply store them as branches.
> See the project branch at http://svn.freebsd.org/base/user/uqs/git_conv/ for extra patches, rules and the scripts used to do these conversions.
--
FreeBSD builds using makefiles, and the BSD make is not the same as GNU make, especially for the sometimes-complicated scripting they have to do inside of them.
Adding cross-compilation to a complicated set of makefiles isn't impossible but can be some work.
Beyond that there are other issues... eg. The ELF binaries generated need to declare themselves as FreeBSD binaries. A C compiler set up to produce Linux binaries won't do that. Maybe also they depend on a specific compiler (FreeBSD is using clang lately). When you build FreeBSD you also build a compiler so maybe this part isn't such a big deal.
NetBSD has had this[0] for ~20 years (requiring only POSIX conformance from the build host). Think of it as a sort of first principle[1] in action. I wouldn’t be surprised if FreeBSD could achieve some success leaning on this work.
Another that the compiler may use platform tools (assembler, linker) to build binaries, and those tools they may need a library (say one for writing ELF binaries) that isn’t available. All of them can be ported, but that can be a lot of work for a limited number of users.
Moreover, those users typically have (access to) the target hardware, as they have to be able to test the end result. So, a ‘normal’ compiler may not be optimal, but still be good enough for them.
Because of that, you tend to see more cross-compilers the slower the target hardware.
If WSL really works out, though, it might be the other way around – Linux on Windows might work really well, be able to do things Linux itself can't, like interfacing with Windows-only hardware, using DirectX, and be the superior experience in total for lots of people. It will be much, much harder to justify the expense of buying a whole different set of hardware and support it if the net result is a worse experience across the board. Businesses tend to not care much about freedom of software and the like.
No, it will not and — without massive changes at Microsoft and in Windows — will never be. Source: have a Windows box, am repeatedly amazed at how bad it is. Seriously, it is embarrassing to use Windows. I think the only reason people enjoy it is the fact that they have not gotten a chance to really spend time in a good tiling window manager on X11/GNU/Linux. The GUI is ugly and laggy. The ads are an insult. The shells are either childish or stupidly (emphasis on the stupid) verbose, and gratuitously incompatible.
There seriously is no comparison: in 2020, Linux is head and shoulders above Windows on the desktop.
>> I think the only reason people enjoy it is the fact that they have not gotten a chance to really spend time in a good tiling window manager on X11/GNU/Linux
I don't like Windows, but its UI is not that bad, come on.
Moreover, I use GNU/Linux full time but I don't use a tiling manager. My windows are fullscreen most of the time, and the rest of the time the features that allow putting windows on the corners or the sides (with the keyboard or the mouse) are more than enough. Tiling window managers are not for everyone.
I didn't downvote but to me the message shows a lack of observation on how people use computers.
The update manager that popups a notification or a dialog box or reboots in the middle of a presentation or a movie or something and, to add insult to injury, the system taking its time to apply the updates during the next startup? that could be embarrassing.
Convenient UX features from the Linux world lacking on Windows are a thing (selection / middle click paste?), but Windows UI seems excellent. It's just less excellent than Plasma obviously, but it's nothing short of bad or embarrasing.
I don't know. I am very OK with Win7 but I tried installing Win10 the other week on a VM in the intent to let it superseed my Linux on the desktop.
The 125% scaling of UI made fonts blurry on my monitor and text inconsistently sized. The "dark" theme lets core apps like Explorer flicker in white then turn dark, like slow Android apps. There are mini ad programs in the start menu. I don't even know where you're supposed to configure settings anymore without search; the control panel vs fancy settings apps vs msc snapins (which are the only parts that still work reliably)?
Their UI is worse than KDE or GNOME now. I had 3rd party applications working perfectly fine which I guess is the point but Windows10 UI is no marvel.
This is a problem because Win7 used to just work, discretely.
What's left is for me to experiment with group policy settings and suchlike (at least my edition permits it) in the hopes of getting my issues fixed. Probably have to throw in a ton of third-party "Windows 10 fix" powershell libraries to clean things up.
I got the Win10 Enterprise license through my company. I would never pay money for this OS, privately.
- DPI scaling turned on by default on high dpi displays even though compatibility isn't perfect. Well, GTK DPI scaling certainly isn't perfect either to say the least.
- Not sure where the settings are: Click "start" and click "settings". Forget about the legacy control panel as that is only there for compatibility with old software and for changing the settings of deprecated features used by old software. Forget about MMC unless you have been directed to use it by the documentation, like for example if you are administering Windows Server machines. And I'm sure it goes without saying that Linux systems are certainly no better when it comes to the diversity of methods for adjusting settings.
And these issues are so offensive that while you thought Win 7 was an "OK" OS, you "would never pay money for" Win 10?
It's true, but just because that makes the Linux desktop dead to me might not make the Linux desktop useless to you.
I did note that the WSL2 push might finally mean Linux gets proper RDP support, thanks to Microsoft. The irony is not lost on me.
When plug-and-play, drivers, and power management work correctly. When business software is compatible. When you don't have to search through obscure forums or jump on FreeNode to figure out why your wireless card doesn't work. When your filesystem corrupts and won't boot because you didn't shut it down correctly. When your font anti-aliasing and resolution scaling doesn't look like shit. When you have several hours a day to set things up, rather than getting work done.
GP did say "in 2020"; I haven't had any of those issues for years
"Works for me, so it's not an issue."
Personally, I run into way more issues when I try to use Windows than Linux; for example, whenever I boot into Windows on my desktop, it doesn't recognize that my ethernet cable is plugged in unless I unplug it and then replug it back in. It also sometimes just doesn't give any audio output when it boots up, and I have to reboot to get it to work. These and other issues I have may not be issues you run into, but they are just as valid experiences as the issues you might have with Linux. My point is that the argument that Windows "just works" out of the box is not as objective as you might think.
I did and switched back to floating. AMA
Not that most people would want to run most of those, but Microsofts corporate customers are fed up with supporting those Linux laptops and all the problems related to that.
I'm genuinely very, very happy developing on Windows with WSL2.
Does Windows set up dual-boot if Linux is installed first? Naw.
Does Windows read or write Linux filesystems? Naw.
Has MS open-sourced any significant or important parts of Windows where it won't benefit directly? Naw.
It is in MS' direct interests if PowerShell or .NET are supported on other platforms; it makes them more credible tools.
Did we get source for Explorer or anything? Naw. Calculator. Big fat hairy deal.
Has MS open-sourced any obsolete products that might actually be of use or interest, such as the final versions of MS-DOS (25 years old) or Windows 9x? Naw. They contain things like Explorer and so on that are still used. Can't go actually _helping_ people.
It's all marketing flim-flam. Don't believe the hype.
Linux is doing great on phones and servers and so on. On the desktop, only ChromeOS is not a rounding error. So, if MS can embrace-and-extend the Linux desktop, and make Win10 a better desktop Linux than Linux, then it helps MS and it helps Win10.
And with what it learns from that, it can apply to WSL in the datacentre and on Azure, where it can make Windows Server a better Kubernetes than Kubernetes.
Which, let's face it, would not be all that hard.
Speaking for myself, I'm coming from the gamedev world (lately webdev) and as such was mostly tied to windows due to various tools(Photoshop,Modo,3dsmax in the past,etc) but that didn't mean that I was going to suffer by using IIS or spend a fortune on Windows server licences or higher hardware reqs for appplication servers for no good reason.
So yes, MS will ignore OSS for their core Windows stuff, HOWEVER they will probably also play nice in the wider OSS (If not FOSS) community since they probably see Azure as the third leg of the future (apart from Windows and Office) and as such will ensure that everyone who uses it (be it for Windows or Linux deployments) are happy with it (And trashing things in the OSS community will thus be counterproductive).
* That they don't care about Free Software (they do, it is a hated enemy).
* That they are ignoring the FOSS world (they're watching it closely and working out how to stealthily attack it).
* That they'll play nice and so should be welcomed (they are a merciless enemy to be constantly wary of).
In other words, their marketing team's message of "we've changed, we're nice and friendly now, trust us" is working perfectly.
Looking at the last few quarterly reports, I'm inclined to agree with the parent commenter here, and your comment doesn't do much to substantiate your point as it stands.
https://en.wikipedia.org/wiki/Embrace,_extend,_and_extinguis...
I was merely agreeing.
Look at MS' history. It used this tactic to kill or maim a long list of companies: Netscape, Lotus, WordPerfect, Aldus, Novell, Sybase, STAC, Central Point Software, Quarterdeck, and dozens more.
It tried it with Sun and Java -- it failed, but only just.
It tried it with the WWW itself with ActiveX and IIS ASP. Again, this only just failed.
It has a long long history of being the nastiest, most deceitful, dishonest, back-stabbing company in the computer industry -- and that is saying a lot.
It is not a coincidence that Gates became the richest man in the world.
Cory Doctorow's Twitter summary is pretty good (short thread): https://twitter.com/doctorow/status/1304811968398729218
The LA Times did a great investigative piece on the Bill & Melinda Gates foundation _thirteen years ago_: https://www.latimes.com/archives/la-xpm-2007-jan-07-na-gates...
TL;DR -- it spends just enough money that Gates & Buffett don't pay tax on the interest of their wealth; it does not do ethical investment of any kind; its objective is whitewashing the reputation of a retired criminal, in which it is very successful.
I have yet to see any evidence that the company culture that Gates created and Ballmer continued has changed in any substantive way.
Netscape defunct 2003, FireFox still exists. Lotus 1-2-3 last release 2002. Sybase still existed until it was sold to SAP in 2010 and SAP killed it in 2014. WordPerfect still exists. Quarterdeck defunct 1998.
Sun, Java, ActiveX, ASP. 1990s.
Anything from this last decade? And, how is "this tactic" going to stop you from using Linux?
> "It has a long long history of being the nastiest, most deceitful, dishonest, back-stabbing company in the computer industry -- and that is saying a lot."
Cruelty, ruthlessness, and exploitation goes with success in capitalism.
> "TL;DR -- it spends just enough money that Gates & Buffett don't pay tax on the interest of their wealth; it does not do ethical investment of any kind; its objective is whitewashing the reputation of a retired criminal, in which it is very successful."
Character attack on a retired man, based on bad things which happened a generation ago, while giving free pass to Google who are strongarming the web with AMP and unilateral changes to Chrome, and unilateral standards which were proprietary Google internals released to the world, because they use Linux in their advertising-spyware mobile platform, and to Amazon, a much maligned company run by one of the current world's richest people, known for its mistreatment of employees and abusive attitude to the publishing industry, and for using its cloud platform to steal and reimplement popular services and put them out of business, free pass for current bad behaviour because they let people run Linux cheaply?
Name another big desktop app vendor will hundreds of millions of seats. Name another industry-leading groupware system.
The last big dominant rival MS took down was Nokia's phones operation, but it mainly succeeded there because Google and Apple accidentally helped: Evidence: MS doesn't make phones any more. It derailed the #1 smartphone vendor's own OS efforts and planted an MS exec in charge so that vendor used exclusively MS phone OSes, but it wasn't good enough. The OS failed and the vendor left the market.
Apple is in a different market. Apple sells hardware. Its OSes are free. It also makes a cut on content for those devices.
Character attacks? Have you bothered to read any of the links I have supplied to support my argument?
But more importantly than all that... MS knows they aren't competing with other companies per-se but rather ecosystems, open source and developer mindshare. If we don't like it we can fork off things and go on our way (like developers did during the Ballmer era despite him jumping around chanting developers....).
Windows isn't even the most used personal OS today, Android is.
Funnily enough thanks to MS in part, not because they sabotaged Nokia (Many N insiders have spoken about how dysfunctional and unfocused their management was even before MS came into the picture), But rather because they fucked over their enthusiastic WinPhone7 users, retailers and devs with WP8 not supporting the prev generation of devices and API's changing.
WP7 had a far snapper and easier to understand UI than contemporary Android devices, but once they dropped the WP8 news the old devices just lay on shelves and sellers who had bet on WP7 selling were left with un-sellable devices and didn't feel like buying a new batch to see it happen again.
I know why the WinPhone fiasco happened: they, very reasonably, wanted to get everything on an NT base. WinPhone 7 was WinCE, WinPhone 8 was NT.
I don't think they realised how many people used WinPhone 7 and were happy with it. Over here in Czechia, Windows phones were the main smartphones I saw when I moved here in 2014 -- Android and iOS were very rare, because they were very expensive. Windows phones were the affordable way to get a smartphone -- there were a lot of candybar dumbphones still around here in 5-6 years ago.
I don't think dodging antitrust was luck, though. I think the US was afraid to kill or even maim one of its very few big success stories, one of the only ones that _wasn't_ being outcompeted by Asian or European companies.
Since then, MS has lost some big battles -- notably, phones and tablets -- and it's learned to be conciliatory and how to appear friendly to people too young to remember the viciously rapacious 1990s MS.
My point is, I still think it is just play-acting friendly. This is the null hypothesis: same company, same people, same structure, and therefore, same business practices. It just got burned a few times so now it wants to come over as a reformed character.
When i mean lucked out i mean that they were lucky enough to have good enough lobbyists to pay their way out of their trouble.
Any company can become aggressive/shitty/evil depending on who runs things. I mean if we look at Suns legacy, Sun today is the biggest threat to open source we have (would would have thought about that in the 90s?).. sure it's really called Oracle now with Larry.E pulling the strings.
MS in the 80s and 90s had lucked in on their DOS deal and i think Gates in the back of his mind might was just watching all the time for threats coming and doing to MS what he did to IBM. But being facing a split-up i think Gates didn't see much point in "winning" any more since doing that would just lead to being big enough to actually be forced to divest things. (So he got bored and started doing other stuff, i know i would)
I think a comparison to the oil industry is apt, early oil was really ruthless in terms of competition (until the breakup of Standard Oil that really MS was the closest to becomming) but todays big-oil industry is quite evil but that is more the making of a collective greed of many middle to high managers and we're seeing the big IT firms approach that state quickly.
... you're fine with Google taking Linux, hiding it inside a proprietary system, not releasing source code for things people want to use like Google services?
> and servers and so on.
A use case where the most desirable setup is to have the Linux servers on someone else's computing platform, preferably in another state or country, with minimal services running on it, preferably locked behind an SSH interface that only an automation tool accesses guarded by an automated deployment tool, trading the "Linux can run for years without reboots" former gloating for being able to kill/redeploy dozens of times per day.
This is what counts as "popular" in 2020, Linux as a small piece of necessary plumbing that everyone wants to push away from human contact. A far cry from the dreams of Linspire, Ubuntu, Ubuntu-phone.
Microsoft make WSL pleasant to install and use, and now it's one of the most talked about ways to deal with Linux.
And you're pretending that what you really wanted was source code for Windows Explorer? Oh please. As if.
I am typing on Linux right now. I work for a Linux vendor. I like Linux and use it every day. I do not run Windows at all, anywhere, unless someone is paying me to.
As for Android, there is this thing called AOSP. Check it out. You might also wish to contribute to Replicant, since you are so passionate: https://replicant.us/
If you look hard enough on Google, you can probably find my (rather positive) reviews of Linspire from back when it was new. (It wasn't as good as Corel LinuxOS.) You'll also find me praising Ubuntu's efforts and bemoaning their cancellation.
Linux is anything but small; it is huge. It is what allows me to keep the lights on and put food and drink on my family's table.
MS executing its hallmark embrace-and-extend on it is _disastrous_ news and is one of the biggest dangers Linux has ever faced.
Don't shoot the messenger.
> "As for Android, there is this thing called AOSP. Check it out."
It's what you get if you remove all the things people want, like Chrome and YouTube and drivers and Google store, and are left with just a kernel which nobody cares about. It's what manufacturers take, and put something nice but proprietary on top before it becomes desirable. (btw, if Android counts as a huge successful use of Linux, how is Microsoft adding a Linux VM option to Windows going to extinguish Android?)
> "MS executing its hallmark embrace-and-extend on it is _disastrous_ news"
It isn't news, WSL has been around for years, and before that the Unix subsystem for more years. And why are you afraid that Microsoft doing literally anything with Linux might lead to you running Windows? How is that going to happen?
> "Linux is anything but small; it is huge."
Then how is Microsoft adding a Linux VM option to Windows a "disaster"? Is it huge and doing great, or tenuous and barely holding on?
> "It is what allows me to keep the lights on and put food and drink on my family's table."
You just said you would use Windows if someone paid you. Linux being extinguished wouldn't stop you from earning. Not to mention the doublethink involved in how Linux can both "be extinguished" and still exist in Windows, or how Linux being used in Android, extended in proprietary ways, is great, but being used in Windows is not great.
Because it's just anti-Microsoft fearmongering, not anything which makes rational or reasonable sense.
EEE was about standard protocols; if Microsoft extend SMTP or HTML and everyone becomes dependent on their version, they could leverage that to kill Netscape. If Microsoft put a Linux VM in Windows, that doesn't change what happens on your server, on your phone, on your laptop. If people start developing for "Microsoft Linux" because it's /better/ because Microsoft builds things people want, that's called "competition".
Linux is strong on servers and phones.
This is, in part, because you can deploy tens of thousands of servers (and tear them down again) at very little cost, and Google gives Android away.
No software licensing.
Being cheap to do means not much money goes into Linux development as a result. Google is doing it to weaken Apple.
Always follow the money.
Windows still dominates desktops. This makes MS money. Not from Windows, but what goes on top: Office and Exchange and Server licences. Evidence: MS gave Win10 away for free in order to get it established.
If Linux development slows and stabilises, then it will be easier to run Linux binaries on other OSes that merely emulate Linux. (Examples: WSL, FreeBSD Linuxulator, Joyent SmartOS.)
If MS manages to make WSL a better Linux than Linux, then it strengthens the market for Windows running Linux workloads, which weakens Linux.
There are multiple markets here. You have to consider all of them and their interactions.
No, E&E is _not_ about standards and protocols. It is about money.
Netscape was free for non-commercial use. In business, you had to pay. Also, Netscape sold one of the leading web servers. That is how Netscape made money.
MS made IE free, but free is no good if the product isn't as good. So it built IE into Windows, and IIS into Windows Server, and gave them away for free, including upgrades. To make them desirable, it added some snazzy interactive stuff in the form of ActiveX (client-side) and ASP (server-side).
Result: Netscape couldn't sell its pro products any more and went broke.
Ubuntu is big on servers because Ubuntu is a good desktop, and when it came out, the other decent desktops -- e.g. SUSE -- cost money. Ubuntu took their market share but didn't make any money. Within a decade, most young techies knew Ubuntu first so they chose it for server stuff too, and servers make Ubuntu money.
Ditto Red Hat and SUSE (now). Give the desktop away for free, make money on servers.
Follow the money.
If Windows can do that better, nobody will bother with Linux. The money goes to MS instead, then RH & SUSE & Ubuntu will die, and I will be out of a job.
It is not about standards. It is about money.
Microsoft has indeed embraced Linux, in the context of Azure, they are making a ton of money thanks to Linux.
And if people develop for Linux-based platforms, why not give them the tools?
Thanks you proved my point
Linux on the developers' desktop, that matters.
That is "Extinguish" is not Microsoft forcing you to stop using Linux it's Microsoft making Windows so good that you no longer want to use Linux.
Linux users think improving things so people like them is bad, because Linux has a culture where you have to suffer the difficulties as a hazing ritual to prove your geek cred, and if you like popular, convenient, easy, pleasant, non-technical, user-or-business focused features that's inherently a "luser" attitude and a person to be scorned.
That is, people who run around the internet fearmongering "Embrace,Extend,Extinguish" are simultaneously criticising Microsoft for making software so bad that nobody wants to use it and losing market share and desperately clutching at popularist straws, while at the same time terrified that Microsoft will make software so good not only will it dominate the market, but they peronsally will find it irresistable and be unable to muster the religious willpower to stick with Linux when faced with the temptation of something better.
It's been 8 years since I've even had to touch Windows professionally and without any overt effort on my part.
I think you overestimate the prevalence of such environments.
That's the context where this matters.
"Oh you're a dev? Here's a mac"
Since 2013 I didn't even have a job anyone with a Windows machine /could have done/, because the stuff simply doesn't run or compile. (Yeah, WSL and all, the box was too slow to compile stuff natively already).
a) it's cheaper b) avoids lock-in and other dependencies
I don't see either of these being much of an advantage for linux on the desktop, although they will continue to be on server and embedded.
It just takes unnecessary effort to keep up with what just works for Linux and I didn't see much benefit sticking to FreeBSD.
1. Ubuntu LTS releases are supported for five years -- Centos is for ten years.
2. FreeBSD updates are typically easy to perform, and the new system is not that different from the previous one. Things are where they used to be and you configure it the same way as you have for years. With Centos, there was a migration from yum -> dnf. With both Ubuntu and Centos the init system has changed from sysv -> upstart -> systemd. The default network configuration on Ubuntu has been replaced with netplan.
I guess what I'm getting at is that there is significantly more churn in Linux distributions and vastly less churn in FreeBSD. This could also be seen as a lack of progress, but I don't think that OP sees it that way.
ifconfig still configures network interfaces (including wireless)
the init system is still the same
you still have the same three firewalls to choose from
the audio system is still the same
The internals are quite a bit different (and more capable), and I don't think zfs was in FreeBSD back then, but UFS2 still works.
If you want to experience twenty year old FreeBSD, to compare with modern FreeBSD, Apple has lovingly preserved it as the userland unix interface in MacOs, they've even got a circa 2000 tcp stack.
What I like with FreeBSD is the slower pace of evolution, so there is less breaking change per unit of time, that makes me spend less time fixing things while being up to date, and when it break there is always a good documentation available. Things I like in FreebSD: pf, zfs, bhyve, rc.d, good separation between user configuration changes and system configuration.
But this isn't FreeBSD development -at least not in the sense that you're creating FreeBSD apps in Linux. This is simply giving an alternative to downloading the isos. If you you need to verify that the source does what it says, now you can build it yourself on Linux.
As someone who dearly wishes they could make NetBSD apps on Linux (edit: I mean build pkgsrc packages for NetBSD on Linux) I don't see it being that much of a game changer -though it is a nice feature on a couple of different levels (being able to build a system on bare metal instead of in a vm, and the code being so portable it can be built in a foreign environment)
From would imply running ssh client on OS A to log into machine with OS B where you run 'make'.
You obviously understood what they meant, as did the rest of us.