Fedora Has Deferred Its Decision On Stopping Modular/Everything i686 Repositories

[phoronix]

Phoronix: Fedora Has Deferred Its Decision On Stopping Modular/Everything i686 Repositories

The recent proposal to drop Fedora's Modular and Everything repositories for the upcoming Fedora 31 release is yet to be decided after it was deferred at this week's Fedora Engineering and Steering Committee (FESCo) meeting...

http://www.phoronix.com/scan.php?page=news_item&px=Fedora-i686-Repo-Fate-Deferred

[M@GOid]

As I said before, Canonical's decision to abandon 32 bit packages will encourage other big name distros to do the same, people crying about it or not. That 32 bit kernel bug with security mitigations going for so long before someone noticing/fixing it, was only ammunition for the argument of ending 32 bit support.

[Weasel]

As I said before, Canonical's decision to abandon 32 bit packages will encourage other big name distros to do the same, people crying about it or not. That 32 bit kernel bug with security mitigations going for so long before someone noticing/fixing it

? Bug reports have been there about it for a long time (since December last year) according to that article. It's not users' fault the devs didn't look into it. You can't make a stupid argument like that.

[Neraxa]

Dropping 32 bit and 686 is a terrible idea and is the distros basically giving desktop users the big middle finger. I urge Ubuntu to start shipping full 32 bit images again and cancel all plans to eliminate 32 bit repos. I urge fedora to continue full support for i686. There are still a lot of perfectly find 32 bit machine around which can be used for instance by schools, libraries, outreach programs to engage people in computer programming and science for intance, all over the country. I do not like to see a lot of hardware wasted when we have a big e-waste problem so it makes use to put old computers to good use. A majoe market that linux needs to support is older computers such as 686, But it can also support the newest and fastest computers as well. I am not against new fast computers and I am for linux being used in data centers, But this does not mean we need to throw users of 32 bit machines under the bus.

I have long been a supporter of Linux for power users and for things like systemd to make linux workwell in data centers. But its not an either or thing. The OS can support both old hardware and new hardware. Im all for new hardware for people who can afford it. Its the people who want to drop old hardware support that are doing something harmful and trying to hurt people who use older hardware because I am all for supporting new hardware, but we also need to support old hardware.

[M@GOid]

Dropping 32 bit and 686 is a terrible idea and is the distros basically giving desktop users the big middle finger. I urge Ubuntu to start shipping full 32 bit images again and cancel all plans to eliminate 32 bit repos. I urge fedora to continue full support for i686. There are still a lot of perfectly find 32 bit machine around which can be used for instance by schools, libraries, outreach programs to engage people in computer programming and science for intance, all over the country. I do not like to see a lot of hardware wasted when we have a big e-waste problem so it makes use to put old computers to good use. A majoe market that linux needs to support is older computers such as 686, But it can also support the newest and fastest computers as well. I am not against new fast computers and I am for linux being used in data centers, But this does not mean we need to throw users of 32 bit machines under the bus.

I have long been a supporter of Linux for power users and for things like systemd to make linux workwell in data centers. But its not an either or thing. The OS can support both old hardware and new hardware. Im all for new hardware for people who can afford it. Its the people who want to drop old hardware support that are doing something harmful and trying to hurt people who use older hardware because I am all for supporting new hardware, but we also need to support old hardware.

A machine old enough to support only 32 bit OS is a annoyance to both users and maintenance personnel. They are slow compared even to smartphones, and there is no spare parts for them on the market. They break a lot (because they're old) and you can only fix them by cannibalizing on machine to fix others, thus reducing the number of available machines to the users. I know this because is what I do.

You don't need to throw them to the trash, there is recycling companies that will buy your discarded ones, obviously for a very low price, but they will not throw them on the trash.

[M@GOid]

? Bug reports have been there about it for a long time (since December last year) according to that article. It's not users' fault the devs didn't look into it. You can't make a stupid argument like that.

The devs don't fixing it is because most of them are paid by companies that don't care about 32 bit. Them if there is no complaining users capable of fixing it, why distros should carry the burden of maintaining 32 bit software? Because the no-paying users (AKA freeloaders, like me and you) demand it?

Lets get real: Linux is what it is today because enterprise money. If it left to the users, you get the situation described above.

[xorbe]

A machine old enough to support only 32 bit OS is a annoyance to both users and maintenance personnel. They are slow compared even to smartphones, and there is no spare parts for them on the market.

This. Claiming that ancient 32-bit machines are great for schools is seriously romanticizing the situation. At least within the USA. Maybe there are other places that people willing to donate endless time to sustain/locate old free hardware due to the inability to secure anything newer.

[rleigh]

A machine old enough to support only 32 bit OS is a annoyance to both users and maintenance personnel.

This is missing the bigger picture. It's not about 32-bit hardware, which has long been supplanted by 64-bit for all but a few special cases. It's about software compatibility.

Removing 32-bit library and loader support from the system is a compatibility break with all 32-bit software reliant upon these facilities. There's a lot of proprietary 32-bit-only code out there. 32-bit ELF binaries have been around, and supported, for well over two decades at this point. There's also a heck of a lot of free software built as 32-bit binaries which many people would like to continue to use.

Systems like Windows, despite all their inherent flaws, made backward compatibility a mantra. Old software had to run on new systems. It matters a great deal. Even today, you can run a 32-bit executable or DLL from the early 90s on today's 64-bit Windows 10. The same line of reasoning also makes retaining such support of high value for Linux, to allow continued use of perfectly functional code. Not all code is provided through distributions, and not all code has a current replacement. I don't think throwing it out because it presents a slight inconvenience is a convincing argument. Automatically building binaries for i386 is pretty cheap to retain, and as one of the people who wrote and maintained the autobuild infrastructure for Debian, which is or was also used by Ubuntu, I'm fully aware of the costs.

[M@GOid]

This is missing the bigger picture. It's not about 32-bit hardware, which has long been supplanted by 64-bit for all but a few special cases. It's about software compatibility.

Removing 32-bit library and loader support from the system is a compatibility break with all 32-bit software reliant upon these facilities. There's a lot of proprietary 32-bit-only code out there. 32-bit ELF binaries have been around, and supported, for well over two decades at this point. There's also a heck of a lot of free software built as 32-bit binaries which many people would like to continue to use.

Systems like Windows, despite all their inherent flaws, made backward compatibility a mantra. Old software had to run on new systems. It matters a great deal. Even today, you can run a 32-bit executable or DLL from the early 90s on today's 64-bit Windows 10. The same line of reasoning also makes retaining such support of high value for Linux, to allow continued use of perfectly functional code. Not all code is provided through distributions, and not all code has a current replacement. I don't think throwing it out because it presents a slight inconvenience is a convincing argument. Automatically building binaries for i386 is pretty cheap to retain, and as one of the people who wrote and maintained the autobuild infrastructure for Debian, which is or was also used by Ubuntu, I'm fully aware of the costs.

I understand your point. Just want to say that Windows retro-compatibility works in theory, but in practice have many flaws. I have see plenty of older software that stop working on newer versions of Windows, for whatever reason. Apparently even 64 bit ones.

And since you are involved with Debian, what is their position on 32 bit ISOs and packages, there is a plan for supporting those until some specific year?

[rleigh]

And since you are involved with Debian, what is their position on 32 bit ISOs and packages, there is a plan for supporting those until some specific year?

I haven't been involved directly with Debian for a few years now, so I can't comment on what their plans are, I'm afraid.