RHEL sources no longer (publicly) available

The existence of a BSD Centos emulator package means the RHEL situation is not off-topic, and in light of CentOS being discontinued eighteen months ago with its end-of-life scheduled for June 2024 means there needs to be some BSD consideration of the linux emulator situation. The knee jerk or natural BSD tendency is to preference the most stable linux environment possible; but that might be a mistake.
What should BSD replace the CentOS 7 emulator environment with?
I admit in the pass I have had a linux copy of office and multimedia editors only because the native BSD version would sometimes suffer a temporary dependency issue after updating ports; I could use the slightly older linux version til the dependency got fixed.
I hope people do not use emulators for daily use, native solutions should be available if you look. I suggest most emulator use is either for low priority evaluation of non-BSD software, a temporary fallback like I was using, or high priority need to interface with some new or nice hardware/firmware that BSD has not had time/resources to address yet (We've all purchased hardware that's unexpectedly turned out to be using some new chip, a problem that is not going to go away). If BSD developers can see how a emulated linux is able or tries to talk to some new hardware/firmware, it should help efforts to port such function to BSD. If BSD supports a older linux environment, it would take additional years or special efforts to learn what the new devices are doing, whereas a linux distribution that works on the newest and widest range of hardware may help BSD developers.
On that basis from what little I've quickly seen looking at linux, Fedora looks like a possibly better option for emulator support.
 
I'm glad I decided to migrate my CentOS 7 server to FreeBSD rather than the new RHEL derivatives now. Of course, it's been a year or two and my migration is still ongoing while the CentOS 7 EOL is only a year away!

Afraid that means I'm going to have to ask some stupid questions here on the forums to hopefully crawl across the finish line - apologies in advance!
 
I feel we at least need a debian one to debootstrap almost a half of distros
And I feel we at least need a fedora/rhel one to yum/dnf bootstrap the other half

For the outliers (alpine, arch, gentoo), their package systems are simple and portable enough to work on the two above. The reverse isn't quite true for apt/rpm unfortunately.
 
BobSlacker said:
Nah, still has apt.
After more thinking, I guess Gentoo would be better. With Gentoo there is no stings attached, we could do what ever we want to.
Click to expand...

My 2c is that I would like to see the distro. with broadest 3rd-party adoption / support targeted. We should not be (in general, and IMHO) looking to enable linux versions of open-source tools that are already available in ports natively, but those things that aren't available — which are frequently closed source / commercial.

See, for example: https://www.mathworks.com/support/requirements/matlab-linux.html. (I haven't tried running Matlab on FreeBSD via the linuxulator, but just a commercial package I know of which actively ships and supports Linux.) They support Ubuntu(Debain) / RHEL / SuSE. If RHEL is off the table due to these changes, my vote would be SuSE.
 
Ogis said:
This always happens when you deal with big companies...just use FreeBSD or Debian.
Click to expand...
Or both...I used to be Fedora guy but got fed up with their almost "continuous integration" release cycle so switched to Debian, so I can enjoy a good solid five years on a release without much hassle. Tried to move to BSD more, but "they keep dragging me back!" with regard to stuff I need that BSD doesn't handle. For now am content to run a BSD VM under debian to stay current on the ecosystem.

As for Redhat...not much good I can write about them. It's pretty obvious why business and guvt choose them, but as we all know, those "business considerations" are often at odds with stuff we geeks value.
 
The thing about RH is that some software is only supported on it. RH is owned by IBM, and judging from comments on slashdot by one person who worked for them, RH always resented that companies would use CentOS or whatever for free. Firstly, when ScientificLinux got popular (people were mad at CentOS because the main developer had the temerity to get married overseas, and therefore, CentOS was late with a point release--IIRC, anyway--and people began turning to ScientficLinux. So, RH quickly hired Scientific's main developer. Then they took over CentOS, I think in 2014 or 15, and as a lot of people know, killed it in 2020 for CentOS stream. So, it seems they're playing a long game. They have seemed to be the Microsoft of Linuxes for a long time now. According to further comments from the one I mentioned, from someone who worked for RH for awhile, they always considered the companies that used CentOS (this was back in 2019) to be freeloaders.
Anyway, IBM has deep pockets and I don't really understand the fine details, but what they're doing is honoring the letter, if not the spirit, of the GPL so I don't think anyone is going to challenge it. Sadly, at least in the US, most sysadmin jobs that aren't Windows, are RH--or up till now at least, RH clones.
 
It was my understanding that the c7 environment provided Linux software a way to talk to the FreeBSD kernel.

Also, since rhel is downstream of fedora wouldn't fedora naturally be the replacement for c7 since the directory structure is identical?
 
kent_dorfman766 said:
Or both...I used to be Fedora guy but got fed up with their almost "continuous integration" release cycle so switched to Debian, so I can enjoy a good solid five years on a release without much hassle. Tried to move to BSD more, but "they keep dragging me back!" with regard to stuff I need that BSD doesn't handle. For now am content to run a BSD VM under debian to stay current on the ecosystem.

As for Redhat...not much good I can write about them. It's pretty obvious why business and guvt choose them, but as we all know, those "business considerations" are often at odds with stuff we geeks value.
Click to expand...
I honestly tried in the past for quite a while (some 10 yrs) to stick with Fedora. Had to abandon it. Same about SuSE. But Debian/Ubuntu is fine.
 
So let's use the original thread. Actually, I couldn't care less about what RHEL is doing – BUT, we DO have an issue with our Linux userland from ports :(

I don't think it really makes sense to maintain multiple userlands in ports, one should be enough. The "straight forward" way now would be to pick some other distribution and re-package that instead of CentOS (and although this sounds simple, it is not ... the userland installed in /compat/linux should integrate and play nicely with the FreeBSD system, and that's not a trivial thing...)

I had another idea in the past: Don't repackage anything, instead bootstrap a GNU toolchain targeting Linux in some "linux-base" port and use that to build existing ports in a Linux variation for installation in /compat/linux. This also sounds simple at first, but is probably even MORE work in practice ... but it would come with the advantage that you could somewhat quickly add e.g. libraries missing from our Linux userland based on existing FreeBSD ports ... :cool: If only I had the time to at least attempt some proof of concept for this ...
 
zirias@ said:
I don't think it really makes sense to maintain multiple userlands in ports, one should be enough.
Click to expand...
The primary reason why all but c6 and c7 disappeared was the herculean effort required to keep all those different linux_base ports up to date.
 
SirDice said:
The primary reason why all but c6 and c7 disappeared was the herculean effort required to keep all those different linux_base ports up to date.
Click to expand...
Which is exactly what I meant 😏 I don't think the maintenance effort ever had a sane relation to the added value of being able to pick a userland...
 
You must log in or register to reply here.
Back
Top