What would you like to see in FreeBSD

Generally, where I've seen hardware issues is with wireless support. I have one Thinkpad L420 at work where I wasn't able to get wireless working, but had to buy an Edimax USB wireless. That worked on 10.x, but not 9.x.

I don't have anything now that's brand new, but wireless is where I've usually seen FreeBSD lag behind Linux. (And if you're paying 21K for something, I'd guess there's a reasonable chance they'd make sure that it's compatible with whatever you want, as opposed to a new $350 laptop.) :)

My UX31E Asus Zenbook didn't work with FreeBSD wireless till 10.x. So, not trying to read anyone's mind, just saying what _I_ think about it, again, since this is a wish list, I'd like to be able to buy any kind of laptop or tablet, including cheap ones, and not have to worry about it working with FreeBSD, just take that as a given. Linux, even if it's got some tiny percent of the desktop, seems to have gotten mainstream enough so that many vendors are making sure they have Linux drivers.

Oko I did a grep -i eth on your dmesg output and didn't see anything. I was too lazy to look harder, but I assume your network cards work.
 
What would you like to see in FreeBSD?

I'd like to see FreeBSD with graphical desktop of standard form, but it is to lose the time, much was discussed of FreeBSD with graphical desktop, and nobody listens. But modern times are giving the reason, the old method is becoming obsolete and will pass into oblivion, the Firewall PF should be configured by default when installing FreeBSD.
 
Oko I did a grep -i eth on your dmesg and didn't see anything. I was too lazy to look harder, but I assume your network cards work.
All those years working with Linux as a paid job apparently affected you more than me :) BSDs use device drivers for network interfaces not eth like Linux. Joking aside igb0 is up and running.That is my second rig with 10 Gigabit Intel and we like it. The controller is quad so one of these days I will play with aggregation.
 
Heh, usually eth will show in a dmesg on most machines I use, so I stand by it. :) That's why I said I was too lazy to delve further.
For example, on a laptop I have (on my lap, as I type on a CentOS desktop) :), I get

re0: Ethernet address
and then the MAC address--above that I get a mention of it with the words Gigabyte Ethernet.

However, that doesn't get my Intel wireless card on said laptop. (Which is why I added I'm too lazy to go through the file and find the card.)
 
I'm working on device drivers in FreeBSD, Particularly on network device drivers. Unfortunately there are few documents for developers who are new to device driver development in FreeBSD (just like myself) while there are many tutorials and documents for developing a linux driver all over the Internet. I think FreeBSD developers must share their knowledge in form of tutorials, especially on device driver development.
 
j4ck - Just check a book store of your likeing and search for a book called "FreeBSD Device Drivers". That should cover most of your needs.
 
  • Thanks
Reactions: Oko
j4ck - Just check a book store of your likeing and search for a book called "FreeBSD Device Drivers". That should cover most of your needs.
Yes, I have bought the book, and I have to say that's a great job, but I think it would be better to learn such subjects through examples. I request FreeBSD device driver developers to share their experiences.
 
I'd like to see FreeBSD with graphical desktop of standard form
Such things are third-party software and not part of the operating system so you're barking up the wrong tree for that. Even Linux uses third-party window managers which aren't part of Linux and only included with the millions of distros of which therein lies a huge problem which FreeBSD, thankfully, is not a part of.
 
Such things are third-party software and not part of the operating system so you're barking up the wrong tree for that. Even Linux uses third-party window managers which aren't part of Linux and only included with the millions of distros of which therein lies a huge problem which FreeBSD, thankfully, is not a part of.

Absolutely agree with that. We don't want FreeBSD defined by the one standard DE choice, whereby all others only get half the love. If you look at the threads asking which DE people use on their FreeBSD desktops, you realise that you'd be pissing off more people than you please by favouring a certain DE. There's so much diversity!
Sure, it could be a little bit easier to get a DE environment up and running and properly configured. But I do like that you have a choice with FreeBSD, which does not require you to clean up a pre-installed DE and resolve a huge mess of dependencies while doing so. An empty box to put items of your choice into is IMHO much better than a dumpster full of unwanted crap. :)

What would I like to see in FreeBSD?

  • improved support for distributed block devices or file systems, ideally with multi-master support, like GlusterFS, DRBD
  • Xen Dom0, bhyve improvements (both actively being worked on, which is great)
  • PF update to more recent version
  • easier way to control power consumption on laptops, including improvements to suspend/resume
  • bsdinstall option that allows for encryption of certain partitions, not only all-or-nothing; can of course be worked around in the fixit shell, but would benefit the less advanced user
  • better support for "foreign" file systems, like NTFS, which would make dual-boot much more attractive to users who also run Windows; IMHO there's currently no working way to mount NTFS via fstab
  • focus less on compatibility with other worlds (apart from file systems):
    • Linux with systemd and Poettering's suggestion to ignore POSIX surely doesn't care about BSD, and trying to play catch-up with them while they are busy distancing themselves from Unix philosophy only takes resources from more important things, and may ultimately either fail or cause some sort of compromise. (Poettering is employed by Red Hat, who do have massive influence on the Linux landscape; nearly all use his PulseAudio and Avahi, and pretty much all mainstream distros are implementing systemd.)
    • Skype and Flash compatibility are lost causes

(This "wishlist" is based on 10-STABLE.)
 
  • improved support for distributed block devices or file systems, ideally with multi-master support, like GlusterFS, DRBD

+1

I would also like support for managing SMB shares directly with ZFS, and per dataset encryption.

edit: curious question, could distributed clustering be done through GEOM? How would one implement such a feature in FreeBSD?
 
+1

I would also like support for managing SMB shares directly with ZFS, and per dataset encryption.

edit: curious question, could distributed clustering be done through GEOM? How would one implement such a feature in FreeBSD?

Last time (3-4 years ago) I tried distributed block devices with FreeBSD (apart from several failed GlusterFS attempts over the years), was by following this: https://wiki.freebsd.org/HAST. It's "just a GEOM provider" as they say there.
It worked okay, but I didn't have enough confidence to run it in production, plus it's master/slave not multi-master.
 
I've played with HAST/CARP for a little bit too recently and it works perfectly for its' purpose (HA/FO), but it doesn't offer a way to aggregate I/O in a parallel fashion, like say, Lustre for example. I would like a way to consolidate resources between N number of nodes and in parallel push it to clients. With FreeBSDs unmatched networking stack this would be perfect, even for "cloudy" stuff. Apparently pNFS (Parallel NFS) does this but I have no idea if it's production ready or if FreeBSD is supporting it.
 
I concur that Linux support can be dropped. It is better to improve applications that run natively.

Updated PF which works with vnet Jail(currently breaks), or,
Updated IPFW, with easier syntax and fully functional built in NAT => https://wiki.freebsd.org/IpfwNg
Jails with VETH, and better resource controls.
Support migration of Jails from one host to another.
Policy on non-free drivers like that of Debian, binary blobs are separate but you have provision to use them.

I have used my experience with iptables, OpenVZ containers to compile this.
 

http://lists.freebsd.org/pipermail/freebsd-hackers/2015-August/048090.html

What is needed to keep sparc64 etc. going is users who have the mentioned machines are also willing to put time into testing and resolving issues, essentially becoming FreeBSD developers for the mentioned platform. The existing developers can't be expected to work on platforms they have no direct access to.

(I think this thread has derailed a long time ago, whatever we write here won't be read by the people the postings are directed to).
 
http://lists.freebsd.org/pipermail/freebsd-hackers/2015-August/048090.html

What is needed to keep sparc64 etc. going is users who have the mentioned machines are also willing to put time into testing and resolving issues, essentially becoming FreeBSD developers for the mentioned platform. The existing developers can't be expected to work on platforms they have no direct access to.

(I think this thread has derailed a long time ago, whatever we write here won't be read by the people the postings are directed to).
Yes, I also think it has derailed. But I didn't create this topic so that users can write their ideas here, but to direct them to topic at reddit, where (I think) developers actually read the posted ideas.
 
users can write their ideas here, but to direct them to topic at reddit, where (I think) developers actually read the posted ideas.
One FreeBSD developer started a reddit thread on their own. There are more developers here, more yet on the mailing lists.
 
The ports tree has been a mess, but they've made a lot of improvements recently. One thing they can do, is encourage through passive changes is, promote BSD, Apache and MIT based dependencies to be used when porting. There can be a make.conf option for those who want to override the default to use L/GNU dependencies instead. It can automatically switch over to what ever is available if a dependency is broken or vulnerable, provided the port maintains compatibility.

Oko -
math/sage replaces Mathematica, math/matlab-installer and other math software.

Replacing Sendmail was always a good idea, and it's a good suggestion to replace it with DragonflyBSD's mail client.

I think FreeBSD should have network/internet clients in it's base-system, not always servers. Then any concerning port can build on top of FreeBSD's client. The base system client, can then point to the preferred program in ports, if a full server is needed.
 
Wow, that brings back memories. I think that was my first non desktop environment window manager. I think I have an ancient and completely out of date page on it. Yup, I do

http://srobb.net/blackbox.html

I just put that for historical interest--I have not used blackbox in over 10 years, but once it was my favorite.

Heh, looking through it, I probably wrote it within a year of XP coming out, back when people were still looking at MS checking home to make sure you weren't using a serial number more than once as still a new and annoying thing.
 
Most duplicate Linux programs of already native FreeBSD programs are better off being dropped. Keep the plugin or other emulation, and improve it to run on top of native FreeBSD programs, with the freed up resources from dropping duplicate Linux versions.

Linux emulation is better off being for when a program becomes newly available to FreeBSD, until it becomes natively ported. emulators/linux-f10 and dependent programs can already go, since that port has been replaced by emulators/linux-c6.

I wonder, what if so many improvements will bring the wrong crowds or attention? Maybe it won't and possibly more good can be done than bad?
 
Back
Top