I’m just curious a curious user. :oops:
From trying recent builds the desktop seems surprisingly feature-complete and much more stable than it was just a few months ago. Networking seems to be M.I.A. though.
Dirty Harry wrote:
I'm just curious a curious user. :oops:From trying recent builds the desktop seems surprisingly feature-complete and much more stable than it was just a few months ago. Networking seems to be M.I.A. though.
I’m curious also - one of my machines has an SiS900 chipset which Haiku “saw” after starting up…so I messed around with “ifconfig” and “route” to see if I could get things rolling without much success…
And apparently "ping" is/was in bad shape 
The networking team has, uhm, NO active developers. 
We’ll have to wait for other tasks to finish, first, or more people to join. We want to port some existing netstack (FreeBSD or DragonFly).
wkornew wrote:
The networking team has, uhm, NO active developers. :( We'll have to wait for other tasks to finish, first, or more people to join. We want to port some existing netstack (FreeBSD or DragonFly).
FreeBSD
Any good reason or are you just in love with FreeBSD? 
Seriously, it’s not an easy decision. Both have pros and cons. None of them would be perfect. FreeBSD is the safe route, but not exciting enough and its threading model is more difficult to understand and more error-prone than that of DragonFly (which is dead-lock free, fast, lightweight, and simple). OTOH, DragonFly could be more risky.
Well, if someone wants to join our efforts: drop us a mail (join the mailing list) or post here! We’ll find you.
wkornew wrote:
Any good reason or are you just in love with FreeBSD? ;)
Actually, just a knee-jerk response. I have used FreeBSD. When it comes to a choice like this one, I have no real opinion I guess :D. But, if DragonFly is lighter, faster, easier, more feature-complete ââ¬â thats the one I would revise my vote to.
ar1000 wrote:
wkornew wrote:Any good reason or are you just in love with FreeBSD? ;)Actually, just a knee-jerk response. I have used FreeBSD. When it comes to a choice like this one, I have no real opinion I guess :D. But, if DragonFly is lighter, faster, easier, more feature-complete ââ¬â thats the one I would revise my vote to.
What I think Waldemar is saying, however, is that FreeBSD is "tried-n-true" with much backing and support. (I would even go so far as to call it "the standard" myself) – it is what many other operating systems base their TCP/IP stack on (Windows NT/2000/XP for example).
From what I gathered on the network team mailing list some time ago, FreeBSD recently started a project to make the stack much more thread-safe and "modular"… is that project looking any good, Waldemar?
One area that would definitely be worth getting right is the network stack… A solid network stack makes all the difference on an OS. If DragonFly is solid, safe, and secure - then by all means, let’s consider it!
umccullough wrote:
What I think Waldemar is saying, however, is that FreeBSD is "tried-n-true" with much backing and support. (I would even go so far as to call it "the standard" myself) -- it is what many other operating systems base their TCP/IP stack on (Windows NT/2000/XP for example).
Yes, it’s less risky. (BTW, I’m not so sure about the Windows thing…after NT they did their own, AFAIK)
umccullough wrote:
From what I gathered on the network team mailing list some time ago, FreeBSD recently started a project to make the stack much more thread-safe and "modular"... is that project looking any good, Waldemar?
Their work will be in FreeBSD 7. Andre Oppermann released a nice document and a short overview about the upcoming speedups and the routing code rewrite (hopefully making it easier to understand). But regarding modularity, I’ve not heard about it, anymore. I must admit that I’m not very informed on this topic, though. I’m doing other Haiku-related work that has less to do with actual programming.
umccullough wrote:
One area that would definitely be worth getting right is the network stack... A solid network stack makes all the difference on an OS. If DragonFly is solid, safe, and secure - then by all means, let's consider it!
DragonFly also means that we’ll rely on a small team. If Matt Dillon (for whatever reason) gives up we’ll have to maintain the netstack on our own. And the FreeBSD team already showed interest in incorporating their work if it becomes successful, so some day we could probably get the same with FreeBSD. Also, FreeBSD might add new features faster than DragonFly. And not to forget, FreeBSD would be easier to port (thus not delaying Haiku as much as DragonFly).
Well, we will have to do a full analysis (used interfaces, amount of work, risk, could we maintain the netstack on our own?, etc.) and then make a decision. Really, I feel much safer with FreeBSD, but probably we’ll have to do our own netstack modifications to make it more desktop-oriented, some day. This depends how much the BSD guys collaborate with us (they were open to this) and how big our modifications would be. If, for any reason, we can’t get around starting our own thing I’d prefer DragonFly because of its simpler threading model.
Most importantly: we need developers! This is not a quick-n-easy one-man project (though, one active developer could do it). Two or three networking developers would be cool. We have big plans, but nobody to finish them.
wkornew wrote:
Most importantly: we need developers! This is not a quick-n-easy one-man project (though, one active developer *could* do it). Two or three networking developers would be cool. We have big plans, but nobody to finish them. :(
I wish I could help, but my development background is primarily Windows apps (C# now-a-days) - and I have very little C/C++ experience. I also don’t know much about netstack internals
I’d love to learn more about it, but I think I’d be pretty useless to you in the meantime.
umccullough wrote:
I wish I could help, but my development background is primarily Windows apps (C# now-a-days) - and I have very little C/C++ experience. I also don't know much about netstack internals :(I’d love to learn more about it, but I think I’d be pretty useless to you in the meantime.
Well, Haiku also needs more professional apps (and drivers). Hopefully, some day someone gets his hands on a Mono port, so C# developers like you can help with new applications. But I don’t have anything against you trying to learn network development.
Allright. What is exactly should be done? I’m a c/c++ developer, know a little about networking and want to help.
BTW, What do you think about creating CIFS server for haiku?
kovdor_develop wrote:
Allright. What is exactly should be done? I'm a c/c++ developer, know a little about networking and want to help. BTW, What do you think about creating CIFS server for haiku?
If you don’t get much response here, post to the network team mailing list (on freelists.org) - i think the developers spend a lot more time using the mailing lists than the forums
Thank you for the offer to help!
I’d prefer people posting to the mailing list (reaching more developers). Here is a little intro:
http://haiku-os.org/wiki/index.php?title=Networking
It still refers to the FreeBSD netstack. We would also like to take a look at DragonFly.
The first step is to analyze which APIs Haiku is missing to get the netstack running (separated by FreeBSD and DragonFly, of course). For example: sysctl, LWKT (which functions?), …
Please post to the mailing list. I’m sure the other developers have something important to say. You can subscribe here:
http://www.freelists.org/list/openbeosnetteam
About CIFS: one step at a time…without a netstack you won’t get very far. But in any case, Haiku must completely integrate into corporate networks (Windows!!!, OS X, Linux).