Hello! Looks like most of people came to a conclusion that now 1st priority is the following:
1. Work out all usergroup.library issues (if any).
2. Add inetd.
3. Add TCP: device.
This is large enough task so i would accept this as a second bounty.
Where is Targhan? Why are these pages not updated? I'd like to see something.

I know that I will be flamed to ground now, but there is a point when enough is enough... You are "accepting" a second bounty ?! You deliver a crappy half ready stack (and I really don`t understand why people are so happy about it) suddenly pushing it to "final"(even if it wasn`t ready) and now you are asking for another bounty to make things a bit more complete (but still missing such essentials like.. err.. well a gui???)?
Well it seems I was right when I thought that guy is just behind the bigger bounties on Aros and Mos.. With that behaviour you simply prove that. Ok, probably I am not allowed to criticize that behaviour either, as I didn`t donate a cent for that stack, but now I am happy I didn`t and I hope no one else will (well I still hope 1.5 comes soon enough and makes this attampt of a tcp stack obsolete)
I know, hard words.. And I know, certanly none will agree with me, but that had to be said, really!

I have to agree the stack is not what it should be. Its slower then the emulated Amiga ones and doesnt have a gui as the the requirments pointed out. It does however do the job and anyone that needs to get online with MOS and doesnt have Miami or Genesis can with it, so its better then nothing I guess. For the ammount of money you got for doing it I had hoped you would put more time into updating it as time passed out of your good will. I guess I was wrong.

As for Targhan I have no idea where he is and I cant update pages for the bounty site still.

I must defend the project. I have been critical of it, because it was
clear from the start that this was based on an old stack and would
never be very good. But since the "official" stack is delayed longer
and longer, I think it is very useful to have *something*. Yes, it's a
bit slow, but this should have been expected from the start.

Personally I have Miami and now switched to MOSNet. It's slower, yes,
but Miami was quite unstable and the connection would regularly go
down, often requiring a reboot to be restarted. MOSNet is much better
in this regard. I still have a problem with MOSNet when there are too
many connections (like when using bittorrent), the connection will
fail then and cannot always be restarted without a reboot.

IMO a GUI would probably be too much work, at least if we don't want
to abandon any hope for the 1.5 stack. I think any future work on
MOSNet should better be to make it work as well as possible with the
things people commonly use it for. I would like:
a) a somewhat better installer and *very* clear instructions
b) fixes for known issues

But I really do hope for the 1.5 stack still so I don't think it's
worth it to spend a huge amount of time eg making it faster or
creating a GUI. Btw, regarding usergroup, it seems to me that it does
not use env:HOME properly (or at least I can't make it work with
bittorrent).

I am using MOSNet too now.
It works quite good.
A GUI is not needed IMO, only a small doc how to set it up would be helpful.
It can be sped up by editing general.conf. Set

TCP_SendSpace=10000
TCP_RecvSpace=10000

or even more like 50000.

(spelling error corrected)

[ Edited by Wishmaster on 2006/4/8 13:57 ]

Well, I'm using MosNet since last xmas so I feel
right to put my words too.

I really don't pretend (and don't need ;) better support
for usergroup.library, inetd and TCP: - a gui perhaps
would be more nice for me, and removing as many linux
dipendences as you can would be nice too.

What worries me is about slowness. I'm not speaking so
much of that 10/20% vs. miamidx emulated when using
internet, I'm rather speaking of SERIOUS slowdowns in
local traffic when using SMBFS. I tried playing with
SENDSPACE and RECVSPACE without success: if I can fix
dld, then upld slows down and vice-versa. A complete
documentation would be HIGLY APPRECIATED.

Also I'm a bit disappointed because I already wrote
you privately months ago without any answer. And I
tried many times to subscribe to the MOSNet ML without
success.

BTW I prefer using MOSNet which is native rather than
MiamiMX emulated, even if many ppl tells me I'm stupid.

just some comments on mosnet:


It works fairly well, but speed could be better, i have no idea what does it but sometimes i get 6-7mb/s and sometimes 2-3mb/s. In a good day, upload speed and download speed is average 5-6 mb/s.(local network)

Sometimes it just stops working, this seems to be random and everything else works(morphos stuff) solution is to reboot. this happends rarely here, but its annoying.

Using rdesktop nearly all the time, i get full systemlock (mouse hangs) from time to time, but i dont know if its rdesktop or mosnet, also very annoying as i cant remember this happening in miamidx.

For another bounty i agree inetd is a possible one, but
fixing usergroup.library and maybe make the gui is a requirement from the previous bounty ? I havent checked but i think a gui is listed in the requirements ?

A seperate bounty for inetd is good, and if some people
are interessted they could dontate, but set some bounty limit, maybe 50-100 dollars is enough ?

but first i agree that the existing quirks should be removed from current mosnet before starting a new bounty.

Wishmaster: Thanks a lot for the tip, this should solve my stability
problems too. These things should definitely be documented a bit
better and in one place!

Anyway I think a new bounty for MOSNet makes little sense.. to me it's
just an "in-between" solution anyway. But I was wrong the last time so
I could be wrong again, of course. I also didn't expect Sonic to
manage a fairly well working port this soon. But anyway.. I would
prefer him to work on some other bounty for some project that can be
useful for a bit longer time.

Sonic

For the amount of money you got for this project that checklist should be completed gratis imho..

It still baffles me that the bounty got that high and baffles me further that it has been cashed out and the stack is worse than the exsisting emu 68k stacks we have had.

I appreciate the work you did, but feel this was an overpaid project. And agree with Acill on some points.

magnetic

I think that
a) It was good to see that the bounty got a fair amount of money in
it, BUT, it's still not that much considering the work involved.
b) When judging what "should be done" for the bounty or not, you'd
better look at the specs for the bounty. No one will ever want to take
up a bounty again if people think they can later include tons of work
that was never specified in advance. As far as I can tell the bounty
was about porting this old AmiTCP version, nothing else. Then people
should not complain about issues that are not the fault of the porter.
c) At least for me, MOSNet works better than Miami, so I don't agree
that emulated stacks work better.
d) As far as I can tell the point of this bounty was not to make a
*better* stack anyway (in that case an AmiTCP port would have been
unrealistic) but to make an *available* stack.
e) So finally, I hope that Sonic will fix any bugs that people manage
to find, but again, please don't try to enlarge the bounty
retroactively.

>TCP_SenSpace=10000
>TCP_RecvSpace=10000

i use (in mosnet:db/general.conf)

TCP_SendSpace=8192
TCP_RecvSpace=8192

you have a spell error too :)


you can also add to samba (samba:lib/smb.conf):

socket options = so_rcvbuf=2048 so_sndbuf=2048

for faster transfers from between mosnet and other
computers on local network.
Without this line it crawls in my network.

smbfs shares are fastest in shell here, you can also
try to play with buffers in shell transfers ie:

copy <smbfs mount> testmovie ram: buffer 10000
copy <smbfs mount> testmovie ram: buffer 100000
copy <smbfs mount> testmovie ram: buffer 1000000

you could see speed differences and find a suitable buffervalue

Ok, as someone corrected me, the bounty specs clearly say there should
be a GUI.

I'm not really sure if this bounty has been paid in full or only
partially (there seems to be different info on different pages).

I'm not sure what is best to do here, but I do feel that spending a
lot of energy to make this stack into the most polished piece of
software available would be a waste of time. It's based on old stuff
and clearly the stack that is being worked on for 1.5 will solve many
of the issues people have with MOSNet. I think bounties are a good
idea basically, but I don't see the point of making people do
duplicate work. MOSNet as it is is good enough to get a basic
connection and that's a huge improvement over nothing. But further
effort in this area should surely be directed towards getting the more
up-to-date and polished stack worked on by the MOS Team.

Gelb: Tried using the log tool? If you got the same problem as me with
bittorrent, it's related to the very low buffers, catohagen's options
should fix it. Didn't notice any enormous CPU loads yet. Resolving
hostnames taking lots of time is due to weird config in resolv.conf,
just comment away the lines and it will be normal.

Personally I didn't expect anything at all, and the current state is
something that has many faults but is definitely usable. That's worth
something even if, after re-reading the bounty specs, I agree that it
is not finished. But I'm not sure how much sense it makes to try
and finish this project as stated in the bounty specs.

I am pretty happy with MOSNet in general. As far as a GUI goes, the MiamiPanel
support is a reasonable way of making the stack simple to use. The additional
proposed features are not really needed for general use. I appreciate
the work Sonic has done on MOSNet so far but would have to say another bounty
doesn't make any sense.

>Hello! Looks like most of people came to a conclusion that now 1st priority is the following:

not yet

from the mosnet bounty requirements :

Must support DHCP, have a MUI or other GUI that includes all necessary classes or use default MorphOS 1.4.5 classes.


1st priority should be the gui, as the bounty isnt complete yet.

I'm running MOSNet and am fairly content with it. The specs AFAIK on the gui were not defined in what it had to do. So bickering over that is mute as there is a gui that works with it. Performance of the stack is sub-par (troughput, and DNS (probably more stuff I don't realize)) and should be corrected before any new bounty is allotted for. I don't have an issue modifing text files for configuration, but a gui that would allow you access to those things would probably help a lot of people.

As far as the whole 1.5 thing goes and having a IP stack, has it be released? Nope. Might as well slate it up there with OS4. Make due with MOSNet or something else, but don't support another bounty until the one is fixed and funtioning as one would expect for an IP stack.

JMHO

Now come on, the stack is as I said better then nothing and it works in most cases. The GUI is not done and IMHO the bounty should not have ever been paid. Making a GUI for another app, heck you cant even call it a GUI its more of a status panel then anything else, is not how it should work.

In any case I cant see how this is a 1.0 release and I am not happy with it. That does ot mean I dont like what Sonic has done. I actually like it a lot and he put in a great job so far. It just need more polish to be a fair solution for what we all put into it.

[ Edited by Acill on 2006/4/7 11:12 ]

[ Edited by Acill on 2006/4/7 11:13 ]

@Mithalas

>I don't have an issue modifing text files for configuration, but a gui that would allow you access to those things would probably help a lot of people.

Me neither, and its not like you need to edit stuff every second day either, but it is stated as a requirement for the bounty so
it should be done before Targhan releases the bounty money.

Sonic knew the requirements before he started.

All of the bounty(1600usd) have been payed without meeting all the requirements, i big mistake by the Bounty Team i would say.

Acill

I was quite surpised the full funds for the bounty were dispersed as clearly the end product appears to be some sort of "beta release" The whole point of the gui was for casual users to be able to config the stack without hacking/scripting or editing text files.

I could live with that if the stack was as robust as it should be. However, given the cpu load probs, bandwidth probs, etc it doesnt feel "finished" to me.

And as far as Gelb goes, considering he does all his apps for free and never got a free peg or anything I guess he has high expectations for ppl getting paid for projects. Which is understandable :)

I commend Sonic on getting it this far, but I still think all the funds should have been given out for a beta product that it is atm.

magnetic


[ Edited by magnetic on 2006/4/7 15:07 ]

My 10 usd worth... (or 9.something thanks to paypal)
An yes, I know it's not a lot, but I don' even have pega/mos, and I'm student at university (= realy limited income).

Originally I gave my donation for coding a brand new native stack cause I though the project looked good (good reputation etc) and there definitely is need for native and free stack for MorphOS.

This old-code-port project was (IIRC?) only secondary opinion.. and I thought it might not be bad either. At this point I don't know what I should feel.. I have no way to check this code myself, so I'll have to trust on those people who's responsible of organizing these bounties for them to keep things under controll.

For other projects.. I'm sorry, but I've been waiting this one to finish before I'll consider donating again.. I know it's not fair for you, but it was the first one and I'd like to see results before next round.

Edit:

Next round => some other project I may fund.. I'm not going to donate more for this until I hear (from sources I consider reliable) it's working Ok with features mentoined on bouty-page


[ Edited by JoannaK on 2006/4/7 22:35 ]

Quote:

I agree and had nothing to do with giving the money out. If I did have any controll over it, it would still be there. Again I am not bashing Sonic as its a great product so far, but not what it should be in 1.0 form.

What a surreal thread, especially the harsh grievances over him getting that money.

His bounty is a fraction of a month's salary. Anybody here (of the complaintants) who can port a TCP/IP stack to MorphOS in two weeks please raise the hand? Most of us will need that much time to just understand the fist half of the book explaining the TCP/IP basics :) If we can program at all on that level. Get real, guys. Bounties, even the biggest, are small donations, not fair payment.

If you'd pay fair and square for work and time, you'd have every right of the world to critisize harshly and get upset if your proud demands are not met to the dot on the i. But you don't. If you don't moderate the tone, you will scare away the idealists and soon have nothing. You are not in a position of strength if you can not pay market rates for work. Essentially, you are beggars: Give little, expect much. How long can that possibly work if you don't make up for it with friendly words?

Now let's do a bit of simple maths, to explain why the bounty system is a donation system and not a payment system that makes you the masters:

Let's say there are ~1200 MOS users and every forth would want to buy a good native stack in the next 24 months. That amounts to about 300 potential sales, which is 3 sales per week. I suppose such a stack could be sold for 25-49 EUR, based on average shareware prices and depending on quality. That puts the potential value between 7500 EUR - 14,700 EUR. To that, add updates (over time, updates contribute more than the inital sale): Let's add one non-free update for 15 EUR at the beginning of the second year. Total theoretical value of his work, in the first 24 months: 19000 EUR (23000 US$).

From a monetary point of view, this developer should go commercial, ie. make a commercial GUI etc. (I assume that the core is GPL or similar and must remain open source?), fix the speed issues and sell.

Dietmar you need to read the thread from post one to this one. Most of us are not mad at him getting the money. In fact I am glad he got it, but he didnt complete the bounty as its writen and should not have got it all. Did he do a good job? Hell yes he did, he did a great job, but the GUI is not done and so in that he shouldnt be here asking to start a round two before the rest of the application is a bit more polished and complete to the way the bounty reads.

As to your last pont he took code from another source that was ported to AROS and had a huge start. Its not like he wrote it all from his head. You need to go read the bounty and then come back and see if you feel the same way. He got a large part paid to him and we all said it was fine for that. Again none of us said he didnt deserve the money. I just feel its not a complete bounty project and should not have been cashed out.

Sonic: I know you are reading this and may feel a bit upset at some posts. Just keep in mind this in know way is an attack on you. Your work is very much needed and we all are glad you do what you do. Keep it up, just please do a bit more on MOSNet before calling it done and moving to another bounty.

[ Edited by Acill on 2006/4/7 16:12 ]

Dietmar: Quote:

But thats exactly the point. I guess no one think that a bounty is eqal to a real salary (at least I don`t) and that a lot of idealism is needed to make such things.

A bounty can never be more than a small donation to say "thank you" (allthough several people donated a small fortune for this particular stack)
However, compared to all the other "bountiests" Sonic imho differs a bit.
It would have been another thing if he had deciced from start to make a new commercial stack, which he would have sold then. If there was a demo then, fine. If it worked as it should people would have bought it, or if it doesn`t fit their need, then not.

But a bounty is as you pointed out no salary...

Ah well.. perhaps I was really to harsh, without any reason. I tried Mosnet several times, it doesn`t fit for me, I didn`t donate, and I won`t probably use Mosnet any time soon.
So sorry if I was indeed to harsh.