ASK OpenBSD Journal: Where is OpenBSD headed?

Ok, it's lame to reply to your own post, but I want to start the ball rolling. I would like to see the soft updates perfected so it becomes the standard for ffs on OpenBSD.

Raw speed improvements would be nice. Not that it is bad now, but faster is better.

Really the thought would be this: FreeBSD is the widest used BSD. Correct? So, what does Free have the Open does not? Of that list, what would Open want to do (all that Free does does not meet up with the mission of Open)?

I know nothing about how OpenBSD handles under extreme load compared to FreeBSD. Some tests comparing the two would be nice. Performance on Open is not bad, but why do we hear so little about it? Everyone makes a big deal about the security of Open (and rightly so), but should Open limit its self to being known for security.

The ports system could use some more work. On free, there is a quick pkg to add when you update the ports tree, you do not have to run -current or -stable. That would help Open as well, users could use the ports, keep them updated and not have to run -current.

And cvs is nice, but cvsup is easier for many people to use. Perhaps having cvsup config files that users could run to update the ports and src tree to current with out having to script it them selfs.

How about thread-safe Xlibs out of the box and finally waving goodbye to ye ole a.out. Cardbus and better laptop sound support would be nice to see as well - I see MANY ppl replacing their linux laptops with a healthier openbsd alternative ;)

I am putting my vote in for Cryto hardware support. How about the Redcreek IPSEC PCI card or
one of the cards made by Crysalis ITS.

I am looking forward to use softupdates when it is finally stable. More documentation would be nice too...

Now that SMP is in the works I think the only stop for really big servers is the memory like 1 or 2 GB (if it hasn't been fixed lately).

And personally I would love SMB Mount. The abbility to mount NT Shares like a "real" filesystem. Cause then I could move compleatly from the ugly Windows world and do all kinds of kewl crontab jobs. But thats maybe more a ports thing.

//Jan J

Hi!

I am probably going to be flamed up and down and sideways for this, but hey, I am a complete newbie, right? =)

What I would like to see, for OpenBSD, is a nice, user-friendly, semi-graphic install software, with reasonable default choices for everything. I'd also like something like "Type 'A' for a server or 'B' for a workstation" type of multiple choice install. Either that or a major "OpenBSD installation FAQ/HOW-TO/man page for complete idiots" -- of course, having both (install+docs) would be insanely great.

Let me explain: I have installed FreeBSD, and its installation procedure is nice, quick and painless. Most Linux distros have got *great* (hand-holding) installer systems -- admittedly, geared toward the average Joe.

OpenBSD, on the other side, has a very unfriendly install -- I have tried three times to install it and I managed to flunk it all 3 times. The last try was the worst, as OpenBSD wiped my entire disk, including all the other partitions I had (I backed up all the data, so nothing important is lost). OK, OK, I know: maybe I am just stupid.

I can't find help in the FAQ, the CD does not comes with a lot of info -- and I am going to acquire a (second-hand) machine just to install OpenBSD on it.

Of course, I KNOW that OpenBSD is not supposed to be for newbies -- it's supposed to be for the GODS of UNIX. But I guess it's time for OpenBSD (and NetBSD as well) to take a look at themselves and decide if they want to be more user|newbie-friendly or not.

This is all the more regrettable that I'd really appreciate the type of security and performance that OpenBSD has to offer...

Just my US$ 0.02...

id like to see a -STABLE CVS tree like freebsd, but i also like had installing (and reading) patches too. you learn alot that way.

for installations on client sites, it could save alot ot maintenace time.

this is the only thing that openbsd really lacks since it would make keeping your box secure that much easier and less error prone.

Tough question. Are we trying to grow OpenBSD's usage/adoption? If so, then it's probably necessary to focus on what the general user wants.
Stuff like softdep, better laptop support, threads, more up-to-date ports, more/better hardware support, better Linux compatibility.

Basically, anything that allows OpenBSD to run as quickly as possible on as many systems. It's almost a necessity for people to adopt an OS on their desktop before they'll look at putting it other places.

From there, the sky's the limit (LFS, crypto hardware, wireless support, etc...)

I personally like OpenBSD the way that it is. It's a system that doesn't crash, has excellent man pages, and I would like to see it stay that way.

I feel that we must resist the desire to seek wider OpenBSD adoption. OpenBSD should not try to be everything for everybody. Keep it small, keep it simple.

1. SMP support
2. Support for my onstream tape drive

I see some people complainning about the install. I like the OpenBSD install. From nothing I was able to install it on my box at work in a couple of hours, including using DHCP, and loading the software through the network, it all went pretyt smoothly. That's a lot better than regular Linux installs.

1. i would love SMB Mount - i know of sharity light but a plain SMB Mount would be better in some cases.
2. i don't need SMP personally, but i can see OpenBSD really needs it.
3. More ports, but not a larger base install
4. XMMS 'nuff said.
5. Keep security the #1 issue. That above all else.
6. I can't think of anything else.

Clustering support.
I don't want to use linux on my cluster.

That'd be wonderful: a 16 node cluster of OpenBSD boxes...*drool*

Greetings,

Since OpenBSD is a secure Unix system I would like to see more security, and this time in a non-Unix way.
What few OSs does (no Unix ones?) is to allow a potentially unsecure program run in a safe/secure way.
See http://www.eros-os.org for some interesting scenarios.
AFAIK the most Unix like OSs that allows for this kind of security is Plan9. And teh first building block needed from Plan9 would be union mounts. These are available for OpenBSD, but with the warning that they are not ready for production. So my number one wish is working union mounts.
To make this security available to 'normal' users (non-root) OpenBSD needs per-process namespaces. Clearly this must be added as a seperate sub-system (what I mean to say is that normal useage should not involve these since they would be totally incompatible with Unix processes name-spaces). One would have to start a special shell or something to get the per-process namespace thing going. Then it would be perfectly ok for a normal user to start doing mount's and things without any risk of endagering security for others.

Finally, the distributed file system of Plan9 is nice and could be integrated, too.

Parallel port ZIP drive support would be nice. It's one of the things stopping me from using OpenBSD entirely...

I'm also waiting for SMP support. I think that'll really make OpenBSD a more commercially viable system.

Hopefully this will not be the equivalent of using gasoline to put out a fire. As far as OS improvements I would like to see better laptop hardware support and the introduction of a moused to allow cutting and pasting in a text mode (If OpenBSD already does this than shame on me) The one thing I think that would greatly enhance OpenBSD is printed documentation. I recognize the online documentation is of excellent quality, however having a manual that explains OpenBSD from the OS's history through installation, network configuration and system administration would be of IMMENSE help. There is something about having a book to page through when you're stuck. As an example the FreeBSD Complete book by Greg Lehey (sp?) really helped me out when I was installing FreeBSD and setting up a firewall.

SMP (I am wasting one processor on my work box)
PAM (This would be nice)

1) SMP
2) large file support

I would like to set up OpenBSD for a demo server exposed to the internet. But our application is very CPU/IO hungry and must be able to process files larger than 2GB. At the moment OpenBSD is not a good reference platform. So I think I am forced to use Solaris or NT (with a OpenBSD firewall/proxy in front).

Id like to see more support for raid hardware such as Compaq Arrays and Mylex arrays.

Best Regards
Kenneth

more hype tshirts..

it's all about the tshirts.

chix dig the tshirts...

I'd like to see vrrp, or hsrp. Baiscly some sort of network based fail over.

I whould want to have the distributed filesystem AFS included (ie: Arla, a free AFS inplementation written by some hackers at KTH and MIT, ftp.stacken.kth.se/pub/arla/) and MIT Kerberos 5 to replace the current KTH Kerberos 4. (Maybe MIT Kerberos 5 and KTH Kerberos 4 can coexist in some way?)

Bandwidth limitation. It would be great.

I think a stable and fast SMP support is the most important thing at the moment.
A 64-bit journaling fs would be nice too ;)

It seems to me that OpenBSD is quite unstable now under very hard loading...
It would be nice to fix it somehow.
Kerberous 5 and a lot of hardware drivers gain OpenBSD to WILD succes IMHO

I am a Systems Admin for a very large corporation. OpenBSD's security is huge step in the right direction, and work on SMP is a must. To add to this shopping list which would allow for the quickest adoption in large scale(midrange) environments you need the following:

Raid 5 Support
Clustering
Journaled File System

These 3 items are hugely needed before it could be considered for taking it's place along side our existing production boxes. I am optimistic that once these things begin to show up so that we have the ability to scale, failover, and recover quickly we will have the makings of a powerhouse OS that could start replacing expensive proprietary system that we currently support. But until their is progress in these areas OpenBSD will not start to take on the likes of the midrange platforms we support now.

On the hardware front I would like to see support for Intel, and added support for Alpha.