Thanks go to Krashan for picking up the bountie for the native TCP/IP stack. He has given an 8 month project time. I would expect more to follow on this from him when he is ready to announce it.


To all: Lets get the donations in for this one! He deserves all the support he can for this one! Its a much needed and a huge project.

[ Edited by Acill on 2005/7/19 4:03 ]

|||

[ Edited by Velcro_SP 01.09.2011 - 16:50 ]

Way to go! I have no doubt in my mind Krashan couldn't do it. Good luck.. hope to see results soon, even in alpha-state.

One pick from the eztcp/ip rules:

Quote:

Why on earth it would need a GUI (at first at least)? IMO the GUI for such an app is so unimportant that I'd not put it as a requirement. A simple "Preferably with a GUI" would suffice.

Secondly, there is a typo in "necessary"

Well, it isn't called ezTCP for nothing you know

Seriously, me thinks a GUI is actually important, at least as bounty requirement. Setting up TCP/IP by editting textfiles (*cough* AmiTCP *cough*) is do-able if you RTFM and have some understanding of how things work....leaving out the majority of users

I can happily do without a GUI in an alpha/beta release (and help testing ), but the full, bounty eligible release should have a GUI.

Hi :)

Well, I know this issue must have already been raised before but...

Why develop another TCP stack, when MOS 1.5 is not that far away and already includes one ? Krashan already works on something very promising - why waste some time on something already done ?

Thanks to the OS4 split, the same things are already being developed twice or more.

I'm OK to donate to the team or to any other project though :)

[ Edited by mahen on 2005/7/14 9:16 ]

[ Edited by mahen on 2005/7/14 9:19 ]

Well, I must admit I started to prepare to this project almost a month ago. I started from reading tons of documentation and RFC-s. I also have now a lot of working code. I wanted to be sure I can make it before I've taken the bounty. I'm writing this stack from scratch for two reasons. We all know MorphOS is a single-user OS with limited memory protection. Many consider it a disadvantage. But it may be taken as an advantage and TCP/IP stack can be much faster than bloated U*ix ported solutions designed for multi-user memory protected environment. The second reason is educational, I want to increase my TCP/IP knowledge with all the details and practice.

So well, what is already done?

1) The main task. There are two approaches to the stack in Amiga-like OS. The first, common one is that the stack is an application. It creates bsdsocket.library dynamically in memory when launched. The second one (used only in RoadShow AFAIK) is that the stack is a library in LIBS:. It creates all processes when opened the first time. I've choosen the first approach, so the network is started by launching "ezTCP" app. This app is a MUI app so it has the stack GUI integrated. The GUI is not done yet, but it will be. Screenshots will follow .

2) Subtasks for interface: IP subtask (handles addressing and packet fragmentation), ARP daemon subtask.

3) The complete library skeleton with all API functions.

4) Syslog facility. ezSysLog is a separate MUI app which can be used to catch syslog() output. It is 100% functional now.

5) API functions related to logging and error reporting like SysLog(), Errno(), SetErrnoPtr(), SocketBaseTagList() (partially), GetDTableSize().

As you can see I've started from logging and errors then I'm building IP layer. The stack is modularized and extendable, so after releasing the bounty version, more features can be added. I will report progress on MorphZone. I also plan to release some beta versions and set up a bugtracker.

Quote:

You believe in MOS 1.5 this year?

Because users don't know when will get MOS 1.5, they 'd like payed for
TCP stack. Better payed for TCP which will be able in 8 mounth, then
wait some (?) years for MOS 1.5....

You are the man!



[ Edited by Zylesea on 2005/7/14 10:07 ]

Krashan : well, I guess you are better informed than us on this point, but I indeed hoped so !

It's all a question of choices
What if the xxx.xxx-MOS-included tcp/ip sucks ?
When my account allows it later this month I'll
put more into the fund for sure

A big thumbs up to Krashan !!!

Great work Krashan I am very pleased with your work and know this will come out as a nice stack. Everyone needs to put a nice donation in for this now. You can see it works!

About MOS 1.5, While it may have a native stack we still have no idea when it will be released. Its been a long time now and its good to have more then one. It makes for better versions.

I decided to tell my opinion about this aswell...

I like the idea of "written-from-scratch" TCP/IP stack, but earlier I just thought that it's better to work on something that isn't done yet, rather than something that is working (but could be done better)

So, it is a nice project. If you think you can do it, go for it!

Also, should we add some programs to requirements list? You know, like "Must work with AmIrc, iBrowse, YAM, openttd and mldonkey" (It's pretty obvious that it should be compatible, but some compatibility requirements could be added to the list)

Quote:

Have no problem with this. AmIRC, IBrowse and YAM are used here everyday, I do not have the latter two, but I'm sure there will be people wanting to help in testing. I can of course install openttd and mldonkey if case of problems needing heavy debug.

Krashan, you are amazing! Not only you show an unseen level of commitment, but you also excel in communication. The Reggae documentation is impressive, and your comments here too!

Go krashan go!

1. Thanks Krashan!

2. Thanks to the donors again!

3. everyone, should donations still be allowed during development time? This is up to you guys, since some other places seem to close it off once accepted. (personally I'de like to leave them open).

4. will be updating amounts here shortly, thanks to Velcro_SP (ezTCP) and BBRV (General Fund again!)

[ Edited by Targhan on 2005/7/14 10:11 ]

Even though I am hopeful of MorphOS 1.5 coming this year, I am a firm believer that an extra stack can be useful. Remember that Envoy uses it's own tcp/ip stack, so we know that you can have multiple stacks for multiple purposes (LAN/INet).

Besides, I would rather have two stacks to choose from in case some applications work better with one vs. the other.

Quote:

I think that it should still be allowed. Some people may feel more confident to put some money in once a developer start to work on the project.

Kicked in some $ on my first bounty.....one day I'll have a Pegasos again

Yep, please keep it open ... so more ppl
can increase the pressure on Krashan

Okay, I've changed the code for the accepted bounties list to include the donate button (when there is one). This way, an accepted bounty can still recieve donations -- at the same time it's not in the "open bounties list" so other coders cannot "steal" the project.

Quote:

Meetoo.. My first donation for MorphOS Bounty

My yesterday activity was socket stuff. I've implemented Socket(), CloseSocket(), socket collector (a routine closing all sockets owned by an app, when this app closes bsdsocket.library). Also background was prepared for socket sharing API like ReleaseSocket(), ReleaseCopyOfSocket() and ObtainSocket().