[Raion]

Hello everyone!

I've been quite busy the last couple weeks with the mouse pad sales but I didn't forget about May 21st. It's been 5 years since Nekochan's demise. 

However I didn't want to dwell on the sadness that such an anniversary might bring and in combination with some good news I wanted to instead announce the next project that is going to be around.

IRIX-32, so named for its basis on kernel and APIs of the last 32-bit compatible IRIX (5.3) is a proposed reverse engineering project to be conducted by a team of developers in the US and the EU.

Purpose: We will reverse engineer the version 5.3 kernel with future goal of producing a fully open source reference implementation. This is the first major step and the delivery will be documentation and reference material to enable effective emulation and driver development for IRIX.

Cost: The cost of the project is set at $6500 USD. We will need to fundraise this, but I am also contributing an additional $1000 USD, plus another thousand from three businesses that still use IRIX and I've a long term relationship with for support and training. Thus the TOTAL cost, including externally sourced funds is a minimum of $8500 USD. There are additional goals we will consider if the funding ends up exceeded. I do not feel comfortable asking tens of thousands up front, so instead we will try to keep this incremental. 

Now I will answer some burning questions:

Why IRIX 5.3?

The primary reason is cost and complexity. It's about 1/3 the size of 6.5.22's kernel. The goal is to see if there is significant financial interest for an open source IRIX and this serves as a way to do that on a small trial scale. Many of the APIs and interfaces used by 5.3 are similar to 6.2 and 6.5 varieties of IRIX. The work here will serve as a future basis for IRIX's FOSS future. 

5.3 has an added benefit, as it contains 4.x compatibility code as well. We will be including that in our scope.

What is the timeframe for the project?

18-24 months is our best estimate. This is subject to change, especially if the scope creeps. 

Why should I be interested? 

Even if you don't use IRIX 5.3 this stands to benefit everybody. Namely, having a reference and documentation will improve hardware support for emulation of various devices, and this will serve at a major stepping stone towards a fully usable, fully FOSS IRIX implementation. This is a substantially cheaper and less dangerous option than trying to reason with large companies like HP Enterprise. 

Should we be worried about patents, infringement or anything?
 
IRIX's patents in the US have expired long ago. Patents have strict time limits and those by now have expired. 

Nothing proposed or offered is in any way a violation of US law. We will not be releasing any of HPE's proprietary code and we will be following all best practices and ethical guidelies. 

What platform will be used to fundraise?

That is undecided at this very moment. I have to talk this over with the team but a follow-up will be made. Until then if you want to invest more than a few hundred in the project and help make this a reality feel free to message me. Any private investments we receive will obviously knock down the price goal, and we will be very transparent on the use of funds.

[jan-jaap]

Nothing proposed or offered is in any way a violation of US law.

IANAL, but I think that's a pretty bold claim.

First of all, you're likely breaking the license agreement if you even install IRIX because you neither "paid applicable license fee", nor "bought an SGI workstation or server" (from the license holder, namely: SGI). Effectively you're starting your effort from an illegal copy of the software. This is a problem because the DMCA states "The procurement of the reverse-engineered product must be through legal means and the person must be the lawful owner of the product."

Then there are restrictions on your right to make copies of the software. Making a copy of the software for the purpose of of reverse engineering is not allowed and has been used to sue. The EFF explains it better: https://www.eff.org/issues/coders/revers...eering-faq

Of course the license explicitly forbids reverse engineering, but so does the law in Europe and the US. There are exceptions (interoperability for example), but unfortunately 'retrocomputing hobbyism' isn't one of them. Commercial motives of "businesses that still use IRIX" certainly aren't.

We will not be releasing any of HPE's proprietary code and we will be following all best practices and ethical guidelies.

That depends on the method used.

Clean Room Reverse Engineering is the only method that has successfully stood up in court (see SEGA vs Accolade). In this case, the disassembly team would have to be all first-owners of SGI systems (in order to satisfy the DMCA). Good luck with that.

But I have the feeling you're more thinking of running the kernel object code through IDA Hex Rays and massage it into something that compiles, a.k.a. Dirty Room Reversing. Legally, unfortunately, this is the equivalent of converting a JPEG into a PNG: you're still creating an unauthorized copy of a copyrighted work.

[Raion]

I appreciate the concern Jan and this was heavily discussed including through a legal counsel. There exist no source leaks as to my knowledge of IRIX 5.3's kernel at this time, you are correct in that we aren't starting from it. Unfortunately the inability of me or anyone we trust to satisfy or procure IRIX IP from HPE means all other channels have been exhausted. The only longshot would be if IRIX is sold to say, Xinuos. And if that happens, good luck at they seem as bad as Santa Cruz Operations/Caldera.

We are maintaining both a physical and electrical air gap between sections which enables the members to maintain a sanitized state. However everything involved in this project thus far is not delivering code. There is no documentation of IRIX that exists to discuss every kernel object and subsystem. If we don't do this, then it'll probably be a very long time before any interest is revived. Of course this could very much fall flat. If nobody's willing to contribute then it's very likely the project will stall and most of us will go on to something else. It's a situation however that I don't see being saved and we are explicitly against abandonment of IRIX. I receive no part of the money since I am primarily just the organizer and spokesperson

[robespierre]

I don't think that writing a full functional specification (which is a necessary step in clean room engineering) for a Unix kernel is even practical for the amount of money described. It is much less than companies spent to reverse-engineer the IBM PC BIOS, for example, yet the system is several orders of magnitude more complex.

Just about everything anyone does with IRIX these days is a copyright violation. From websites that host documentation or man pages, to hobbyists burning copies of media to install on their SGIs, the whole ecosystem is unlicensed. If the shrinkwrap license posted above is legally enforceable, then there is no "right to use" simply because you own hardware. The license is nontransferable unless you signed a contract with the original purchaser to assume all of the responsibilities of the license (I don't think anyone ever did that). Note however that 1998 is the date of IRIX 6.5 and the license for 5.3 was almost certainly much different (jurisprudence around shrinkwrap licenses was highly uncertain through at least 1996).

[Raion]

> I don't think that writing a full functional specification (which is a necessary step in clean room engineering) for a Unix kernel is even practical for the amount of money described. It is much less than companies spent to reverse-engineer the IBM PC BIOS, for example, yet the system is several orders of magnitude more complex.

The people involved are aware that they can't make money doing this in terms of time and finances. The purpose is an amount that will give the hard-working people behind it some amount of money for their trouble which is the point. I myself am not receiving any part of the money. My only real job at the moment is representation and organization. I receive zero compensation from any of this.

[jan-jaap]

If the shrinkwrap license posted above is legally enforceable, then there is no "right to use" simply because you own hardware. The license is nontransferable unless you signed a contract with the original purchaser to assume all of the responsibilities of the license (I don't think anyone ever did that). Note however that 1998 is the date of IRIX 6.5 and the license for 5.3 was almost certainly much different (jurisprudence around shrinkwrap licenses was highly uncertain through at least 1996).

At least in the EU the resale of software is legally allowed. Any clause in a license agreement that tries to restrict that is not enforceable in the EU.

Here's the license from the IRIX 5.2 Developer Update from 07/94. Sorry about the pattern in the background. It's not a shadow, it's SGI's funky design.



The bigger problem here is not some a shrinkwrap license says about reverse engineering and decompiling, but what the law says. In both the US and the EU reverse engineering is prohibited except for a few narrowly defined exceptions. In the US, these exceptions have to be re-approved every couple of years, I believe.

[Raion]

Because IRIX CDs are neither serialized individually or have serial keys or other software based restrictions, they can essentially be resold in perpetuity. It's not illegal to resell console or PC games within the United States for this reason because they are not serialized, otherwise GameStop would be a criminal racket.

[vishnu]

Just my opinion but the likelihood of HPE expending any legal effort against anything anyone does with or about Irix is pretty much zero. They might have one of their clerks load up the "Cease and Desist" cannon but I think that would be pretty much it.

[Raion]

For any visitors from tedium.co, yes I am the person who was interviewed and no, I'm not what my avatar shows. I like exploitable manga images, and this was one I've stuck with for a whole.

I invite anybody who wants to join and participate on this discussion to do so. Getting an outside perspective is always appreciated. Let's hear it.

[mxtool]

I'm not a lawyer and won't pretend to speak legalese, but I support this project and would chip in a few dollars.

I agree with vishnu's sentiment, why would HPE care about hobbyists using IRIX software from the mid-1990s? HPE surely can't claim this would in any way damage their business or stifle sales. Heck, maybe HPE can be a sponsor or would want to recruit anyone talented enough to be working on this effort.