numpf.pb

(This should answer your in-game question, and explain the state of PB and other zones to anyone who's wondering) There are 2 parts to getting access to the SSC biller. 1) Adding a rule to the ssc machine's firewall to allow traffic from an IP 2) Getting login id ("slot") so the zone can login to the biller program itself My understanding is that only PriitK can do (1). When powerball knew we were moving we asked for (1) on our new machine sometime in Aug. or Sept. PriitK has not done it, nor, afaik, acknowledged the request. I guess we haven't checked in the last couple weeks though. So, it's unclear whether any new machine will ever get (1). (2) is relatively easy. Any existing approved server can ask for a new zone slot. In the process of changing machines, I created some iptables forwarding rules so that we could route biller traffic through our old machine that already had (1). This allows all SSC billing functionality, but because the client and biller don't see the same IP for the server, the client sends passwords plaintext to the server and you get the GUI warning box. It was meant to be temporary for testing, but because priitk never replied, we (with permission from ghost/polix) set up the same rules on HZ's machine, so now we forward our billing traffic through them. We host SSCX Devastation, which does the same thing, and I was recently asked to set up TW's asss Dev zone with the same trick on their amazon vps. Due to the password issue, I don't think new hosts would be permitted to do this. -numpf

(hakaku replied while I was writing. there is duped info, but I'll post in full anyway) It's a bit hard to make this simple but: security.so contains the encryption and security checksum code. Continuum physics are "client-authenticated," that means that what the client says happens is what happens for the player using that client. If you dont see a bullet hit you, you don't take damage. If you can modify the client's memory or the packets, you can cheat. So encryption keeps you from modifying packets, and checksums keep you from modifying the exe. You need security.so with asss if you want to prevent cheating. If you have its source code you can circumvent it. Its source code needs to remain secret. Grelminar wrote asss originally and was given code by priitk to make security.so. Grel more or less retired from asss dev years ago and gave access to people who were active, trusted, and capable at the time. That's Dr Brain, me, and stag shot. Arnk Dylie also has access, but I don't know if he was given access by grel or brain. Snrrrub has in some ways behaved like Catid. I don't know about the incident Hakaku is talking about, but he has publicly posted some details about cont's encryption and checksums. Which need to remain secret to prevent cheating. I wouldn't give anyone src to security.so w/o knowing and trusting them, and even then I'd prefer to have approval from priitk or grel. Snrrrub can never qualify, for me, based on his actions. I was not asked at the time though; brain can answer that if he wants, but I assume his answer is similar. As an aside, Jowie intends to circumvent all this in his client/server with server-authenticated physics. The server tells you if a bullet hit you. I don't know what other clients like discretion do. -numpf

I see no way of answering your first question beyond what I have said. What might happen to deva if something weird happens sometime in the future? Shrug. We can fix problems with security.so, our host has been around for a multiple of the amount of time snrrrub has had ssca up, and our host is not developing a facebook competitor to cont. That's the best I can do for you. I don't know what stag wants, which is why I suggested he clarify. Peering is only coincidentally offline in PB (and therefore all the zones that use PB as a hub for peering). I have set up a tunneling arrangement, but it would require extra effort to set up peers and make a mess in our firewall config, so we decided to just turn it off for now. I was under the impression stag wanted to put it back as-is once priitk adds us to the firewall, but now he seems to be saying something different. -numpf

Snrrrub changed to a 64bit machine and didn't have a compatible security.so. I have been told he was sent one (or pointed to one) that should have worked, but insisted on being given the src code. It's somewhat plausible that he really did need the src, but given certain concerning behavior of his, and the sensitive nature of the security.so src, he was denied. We don't have the same problem because I have access to the src, and have compiled it for our machine (also 64bit). Deva's staff seem pleased with the setup. They have an ssh account and a couple devs who can do almost anything they need to to maintain their zone. If it really matters to you or anyone, I suggest you wait and ask them how they feel about us as hosts after they've been on our machine for a while. -numpf

See https://bitbucket.org/grelminar/asss/issue/118/threading-locking First, you need to clarify whether you intend to return to a situation in which our pop (or our peers) is inflated. Second, since I'll be handling hosted zones, please don't volunteer me for more work, especially yet If anyone wants to be hosted by PB's new machine, there are conditions. You will have to use bleeding-edge asss, testing changes to it as I make them. All your source will need to be on the server, and I will have access to it, to debug if necessary. And you will have to wait. The way I have deva set up is sloppy and I want to fix it first, so I have a good template to work from. -numpf

crosspost in case people miss it: http://www.subspace.co/topic/26155-ssca-down/#entry283940

the "SSCA Devastation" in the directory list is not actually on SSCA. Powerball has a new host machine and will be hosting devastation. I wanted to leave the name for now because that's what cont uses to store zone files, so this avoids a little extra downloading while we settle on the new host. I guess that time has passed, and it never mattered that much, so I have changed the name to SSCX Devastation The zone is actually connected to the SSC biller. The biller has a firewall that needs to be updated, so for now I am tunneling through our old host. Unfortunately this means continuum sees a different IP than the biller for the zone, so you get the warning box. Otherwise biller functionality all works; your chats will work, and you have to use your SSC password to login. Polix and ghost have been notified of this setup, and polix has accepted it. There is no way around this without Priitk personally updating the firewall. He was notified weeks ago, and pinged again a few days ago. Your guess is as good as mine as to when he'll get to it. -numpf

I saw that, and what's interesting is that he does not mention functions. Strictly speaking I was wrong, a mere header is not enough. Once it declares functions though, those have to be linked to. What I was remembering then is a rule of thumb, mb. It is arguable whether the particular loading arrangement in asss causes binary modules to be covered by the GPL. So, you are safe if the header you include only has those things RMS mentions in it. I would still assume you have to provide source if you write an asss module and distribute the binary. Cheese, are you going to give write access to your repo?

_This_ is why programs ported to linux need not necessarily be GPL: http://www.gnu.org/licenses/old-licenses/gpl-2.0-faq.html#PortProgramToGL I see no definitive answer from a search either way. I can't find the definitive statement I thought I remembered from the FSF. It's an ongoing argument with no apparent court decision to cite. It is possible you are right. I would still _assume_ my definition for safety, use a GPL wrapper + binary blob if you want to get around this, and consult a lawyer if this really matters to you for something else. I overlooked that in the asss license, some files have been placed in the public domain, including util.* This means whether your modules must be GPL boils down to the argument above.

I haven't really used github, but I think the main difference is mercurial support (native on bitbucket, weird on github) and the main difference with mercurial is better windows support. There are other differences that I think go in favor of mercurial, but check here if you like: http://stackoverflow.com/questions/35837/what-is-the-difference-between-mercurial-and-git I am biased, but I'd prefer bb/hg. The 5 user limit only applies to private projects on bb. And you can accept changes by having others fork then make pull requests.

No, I guarantee I am right. If you insist, ask in an expert forum like stackexchange. Describe the situation: You will be told yes and yes. The situation is almost perfectly analogous to linux kernel modules. There's no amount of nuance with asss modules merely calling through function pointers in interfaces that avoids the GPL as long as you are using #include with GPL'd headers (and you still link directly to util.c functions). For anyone who ever finds this mattering for something important, you need to talk to a lawyer. Please do. I see a bug I'd post an issue for or fix: if you use ?fai p=name you will get a deadlock sometime later. The code returns in the middle of fuzzyfind without calling pd->Unlock(). I also see what's causing the deadlock that made you comment the locks in fakemove.c:addfake. The traceback would look like fakemove.c:mlfunc(movelock) -> DOCBS TargetReached -> fakeai.c:TargetReached -> fm->SetMoveTarget -> fakemove.c:addfake(movelock) So the same thread is trying to get the movelock further down the call stack. This can be fixed by making the mutex recursive in MM_LOAD, like e.g. core/cmdman.c does. Given the way asss is structured, you should make recursive any mutex that is held while you call something in some other module (including doing callbacks). Despite your test, this code shouldn't be considered stable until it's tested in active zones with many real players creating lots of player events (especially leave) and arenas being created and destroyed.

Try logging out and clicking it. It's a forum access issue, and you probably have privs as the uploader. I get Doesn't matter. #include of a GPL'd header causes the resulting binary from a compile to be GPL-covered (if also distributed), period. The way people get around this is to compile a binary that makes no reference to functions in a GPL header in any way, then make a GPL'd wrapper that loads the binary and forwards the function calls. Then you still have to distribute the binary _separately_ from its GPL wrapper. Grel didn't or doesn't understand the GPL, though it almost doesn't matter given that he won't bother asserting his rights. It gets messier when you consider that anyone that made big contributions to asss (mgb, catid, akd, brain, others?) might hold the copyright on included code and could assert _their_ rights even if grel doesn't. Anyone writing an asss module should assume they will have to provide source _if they distribute_ their module. All reasons that I don't like and avoid using the GPL

download link is broken. Also FYI, mostly for said aspiring asss devs, asss is covered by the GPL. By #including files from asss, your code is also required to be covered by the GPL. You can't distribute binaries without the source, as you have been, without being in violation of the GPL. Technically grel could send you a cease and desist or something. It also makes your module almost worthless. see http://forums.minegoboom.com/viewtopic.php?p=83700#83228

A url would be nice. It wasn't discussed in game with me, and I registered here just for this, because stag fwded me ssc emails. Simple searches aren't turning up anything obvious right away. edit: in any event, the question is about priit's attitude more than the src itself.

That is your interpretation of PB's intent. I did not mean to say you don't like peering, but are annoyed at the side effect. Your notion of causation is also different than mine then. e.g. website owners who leverage their control over the installer to get their way. People may have misunderstood this, but there has been no SSC vote on anything to do with this issue. It is a unilateral threat by polix. Perhaps he suggested a vote on it, but not in the emails I saw. I can't tell who might misunderstand or who might be misrepresenting with this statement. It certainly doesn't apply, at all, to fix.dll which is a hook into subgame and handles the peering we are discussing. As for cont, all of ek's code is for the GUI/chat window and isolated in menu40.dll. They only share a header file defining the interface. I'm fairly certain the .exe can be modified independently of the dll, provided cont doesn't change the interface to the GUI. I know the exe checksums the dll, so the reverse is not true. Are you serious, in what you seem to be implying, that priit's attitude is that he would hand off the source to someone capable if he had the opportunity? Childish. "Maybe I will, maybe I won't." Sounds like exactly what zippy meant. It might be interesting to read that if it can be dug up. In any event, stag cares more about the peering issue than I do, so he will be making the decision for PB. -numpf