[Raion]

Someone asked me on discord about "Why not Freenet, I2P, Tor, or develop a similar protocol, or use Gopher?"

There's a few good reasons I don't want to do that. For the first three, they're widely used by criminals. I don't want to create an enclave only for it to be a haven for criminals dealing drugs, CP, etc.

Developing a new protocol reminds me of the XKCD thing:



Gopher is alright, but the idea is a low barrier to access. You stumble upon our search engine and mission statements, and then start searching and using it. That's the idea.

[Raion]

I was thinking about the three big things that such a service would need to be successful:

Number one is a news page similar to Drudge Report but that adds the addition of news scraping to be able to view the links on the front page as plain HTML from your computer

Number two is some type of basic social network. In order not to overlap too much with the functions of a forum I was thinking something like Twitter but in basic HTML.

Number three is a video sharing platform. You would have the option to download the video and play it locally on a variety of formats, or you could just stream it if your operating system supports such a thing.

I haven't forgotten about this and I think that all three of those things could be easily implemented but I might need some cooperation. I think using archive.is and a few other free tools plus RSS would make a very decent number one. In particular, I'd focus on non-sensationalist news sources

[weblacky]

I'm afraid I don't grab the full context of the proposal but I can speak to a few things (play devil's advocate on a few):

1. I dearly want websites to go back to simple, fast, layouts that don't need a Ghz CPU to render correctly! So yes, old netscape capable rendering. We don't need all this wasted CPU cycles for informational pages. However, I don't know much about recent HTML standards and fallbacks. I'd never tell anyone to use old standards to start something new. If there is a way to practice good-faith HTML5 with proper fall-back to HTML standards that work with the old netscape navigators for simple or older platforms, I'd like that.

2. Video Media, I'd recommend a stub or placeholder, but I'd not do videos if I were you. Storage and bandwidth requirements aside, I've not yet seen a good use for them for anything OTHER than repair HOWTOs. I'd leave the video storage to YouTube and perhaps just take the videos out of public search?

3. A special protocol...why? I even asked about a paywall before. What's really the difference? Really it's the location of the front door. It's no different from the issues on an adult entertainment site or a bank site. Put up a good login page and https the rest. Why would we not use normal HTTP with TCP/IP? Work on getting a great, fast core, site running.

4. Learning from nekochan...another way? I'm old enough to have use usenet groups in the early 90's. I'd kind of bite on that, you want clean, downloadable, and reto...why not run a secure NNTP server backend? This way duplication is crystal and easy. You can place a web-front end for basically a newsgroup backend. We could use newsgroup applications on our platforms (no web browser needed) to download old and new discussions. it also means every user could be a site archive?

5. Invite-only, I'm on the fence (though I made the same suggestion) because I don't actually know anyone here personally. I've been a long-time contact of Ian at SGI Depot, so I guess I knew one person that could have invited me...but really, no one. I'd say we actually do something more sideways.

Register SGI systems!

That's right, let SGI be the door. If you own an SGI system, you can register it. We build up a HUGE registration list of email/URL/SKYPE/whatever contact info for people who still own systems (maybe you could produce an awesome inventory binary that we can run on IRIX that would inventory the systems and auto-post the system info to our user account info!). We could also ask that the systems be registered as broken, functional, etc.. Perhaps get pictures too! Similar to SGI Supportfolio process...you can could even run the front-end netscape-capable website for the interface? First-time registrations can add text of the story of where they got them, original pricing, who they think used to own them (business name), dates of purchase, etc.

Now why? Because these system are traded, sold, stolen, parted out? Why not have system ownership be the initial key? I'm unsure if anyone ever decoded the SYSIDs? Anyway, the point is SGI newsgroups are for SGI owners.

When a system is transferred to a new owner, the new owner can try to register it, if the old owner is known to the site it will flag them, and try to send the old owner a notice to please confirm the sale of the system. If the old owner isn't responsive in like 30 days, then the new user is let in under an "unverified" system, and it's tracked, if the old owner later appears they can validate the previous transaction. Also, if the old owner can produce a "sales receipt" code for the system, then no confirmation from the old owner is needed. The new owner is fully validated by having the receipt sales code.

This would also give you a searchable list of users that have the system type you're interested in, in case you want to target messages for parts or systems for sale.

Anyway, I think this at least "a" barrier to just anyone. For those people that are interested and looking to get involved, you'd need a guest mode to view but not post in most areas. Perhaps I "newbie" area where guest registered users can post, they become verified users with the registration of an SGI system! Also the marketplace would be enhanced by selling registered systems and that way the users have history, condition photos, and a trail of events.

Lastely, something I personally want, is a custom remote OS loading subsystem (roboinst?). That is, a system where one or more people (acting as a single, unified group), offer a VPN system and hook up (via your home router or configure a Linux VM under hyper-v or something). To pay the group for a week-long pass, to log into the VPN and have the VM or Router provide an IP address to a booting SGI system and all for remote OS loading, even custom images like Ian does. I have several SGIs (a lot) that need reloading after fixing. If I could pay either per-time or per system. I'd do like $15 a system or $40 for a week and hook up my SGIs to the internet and start remote loading. We know SGIs don't go very fast anyway...many of us have at least 100Mb Internet connections or better. So loading remotely is no more slower than locally.

If i could pay Ian to have his system-special images...I would (without sending a hard drive or something). If you could produce a system like that, software distribution would take off!! Also, we could do things like ask Ian for a copy for his device driver library and perhaps have installation options for known add-ons and such (made easy). I'd love an auto-detect script that runs and finds what hardware it can, and runs the driver install packages from the SGI library for me. Even getting a fully working system isn't too easy with SGIs. RoboInst scripts can can do a lot of this and some variables for the system type even present for scripts to use. Id I had the time, I'd look into this big-time to create an autoloading network. Pay to play...

[Raion]

I'm afraid I don't grab the full context of the proposal but I can speak to a few things (play devil's advocate on a few):

This is not an SGI focused proposal, I fear you're confusing this with other posts I've made, but I'll advise as best I can on the relevant posts.

TL;DR: I want to create an HTTP-only small enclave of the modern internet that is independent of any services provided by Google or other big companies. That includes refusing to have it indexed by Google and other search engines - we'll have our own engine and a word-of-mouth marketing. This is to protect us from the encroaching toxic politicization groups and "normies" like your crazy uncle or your millennial niece who virtue signals constantly.

1. I dearly want websites to go back to simple, fast, layouts that don't need a Ghz CPU to render correctly!  So yes, old netscape capable rendering.  We don't need all this wasted CPU cycles for informational pages.  However, I don't know much about recent HTML standards and fallbacks.  I'd never tell anyone to use old standards to start something new.  If there is a way to practice good-faith HTML5 with proper fall-back to HTML standards that work with the old netscape navigators for simple or older platforms, I'd like that.

Web standards have changed, and many aspects of pre-HTML5 pages are deprecated or even not supported, such as the <blink> and <marquee> tags. There's little forward compatibility built into HTML5, so the answer is to stick to HTML4.x compliant techniques if you want fast pages.

Irixnet's current pages are as basic as they can be while still being html5 compliant; I have been working with our webdev to develop an HTTP gateway. Probably would look something like this: http://irix.cc/scratchpad.html and it'd enable basic HTML/HTTP access.

2. Video Media, I'd recommend a stub or placeholder, but I'd not do videos if I were you.  Storage and bandwidth requirements aside, I've not yet seen a good use for them for anything OTHER than repair HOWTOs.  I'd leave the video storage to YouTube and perhaps just take the videos out of public search? 

I want a small scale service for a service to share videos, animations etc. This is not meant to replace youtube, and I simply cannot scrape them from Youtube. There's a mire of TOS, and conversion of Youtube based videos is not easily possible. You can get sued by Youtube easily - see what became of the old Hooktube for reference.

3. A special protocol...why?  I even asked about a paywall before.  What's really the difference?  Really it's the location of the front door.  It's no different from the issues on an adult entertainment site or a bank site.  Put up a good login page and https the rest.  Why would we not use normal HTTP with TCP/IP?  Work on getting a great, fast core, site running.

I pushed back on this when people suggested. You can't do freenet or anything for this to work.

4. Learning from nekochan...another way?  I'm old enough to have use usenet groups in the early 90's.  I'd kind of bite on that, you want clean, downloadable, and reto...why not run a secure NNTP server backend?  This way duplication is crystal and easy.  You can place a web-front end for basically a newsgroup backend.  We could use newsgroup applications on our platforms (no web browser needed) to download old and new discussions. it also means every user could be a site archive?

We have RSS, but that's not 2-way. I cannot recommend NNTP because it's absolutely a bad, insecure protocol. The SGUG guys keep talking about NNTP, but at worst, that'd get like 10 users. It's not going to happen. In reference to IRIXNet specifically, I want a plain HTTP gateway as I said.

But larger scale, it has to be HTTP-only (no SSL, because SSL can be canceled and isn't supported by most old browsers) and it can't be indexed to protect it from toxic normies.

Register SGI systems!

That's right, let SGI be the door.  If you own an SGI system, you can register it.  We build up a HUGE registration list of email/URL/SKYPE/whatever contact info for people who still own systems (maybe you could produce an awesome inventory binary that we can run on IRIX that would inventory the systems and auto-post the system info to our user account info!).  We could also ask that the systems be registered as broken, functional, etc..  Perhaps get pictures too!  Similar to SGI Supportfolio process...you can could even run the front-end netscape-capable website for the interface?  First-time registrations can add text of the story of where they got them, original pricing, who they think used to own them (business name), dates of purchase, etc.

Few problems with this:

1. I don't want to exclude users from parts of the site by virtue of them not having a machine.

2. I don't think any of us should be contacted directly in lieu of the forums, but there was a concept at one time for a parts registry, system registry and more sophisticated marketplace. That evaporated because the person behind it fell through on it. If he wants to talk about it he can here, but I'm not about to drag him into the discussion on it. He and I have had discussions on it, but he has too many reservations to make it here anymore. I suspect it's partially because I'm a main admin here. Me stepping out/down is not ever in the cards. If that means we miss out on these concepts, fine, I have no issue funding/building my own, but I'm not a man who can afford to do that by myself.

3. We have silicon image, which functions as an image database. It's not been heavily marked/advertised because it's not trivial to add it to the pages here. A lot of templates would need to be redone, and I'm not willing to do that until MyBB 1.9, which will require a redo of templates anyways, comes out.

4. There's a few patterns to S/Ns, but some of them are generic, indicating the model type/config it came with, I don't think all of them have decodable asset tags. No

Lastely, something I personally want, is a custom remote OS loading subsystem (roboinst?).  That is, a system where one or more people (acting as a single, unified group), offer a VPN system and hook up (via your home router or configure a Linux VM under hyper-v or something).  To pay the group for a week-long pass, to log into the VPN and have the VM or Router provide an IP address to a booting SGI system and all for remote OS loading, even custom images like Ian does.  I have several SGIs (a lot) that need reloading after fixing.  If I could pay either per-time or per system.  I'd do like $15 a system or $40 for a week and hook up my SGIs to the internet and start remote loading.  We know SGIs don't go very fast anyway...many of us have at least 100Mb Internet connections or better.  So loading remotely is no more slower than locally.

No, this won't work. It requires root RSH, TFTP, bootp() etc. and that is a massive security risk. It can't be done safely even with a VPN, which is why I have custom scripts to turn on and off those services as there's demands. It's not going to easily function.

There's some hope if someone fulfills the bounty for the inst replacement we could embed ssh and other services, but that still would require a local bootp(), tftp, etc.

From a technical standpoint, bootp() interferes with DHCP and vice-versa, and I find that I have to use only static IPs on the network to use bootp(). In short: Not a serious option for something. There's booterizer, but I'm not fond of that project because the progenitor of it has decided to attack me for my community role in a rather insulting message. The fact remains, these kinds of projects are more trouble than they're worth and may have at most 4-10 users either due to scale, practicality etc.

System image-wise, that's actually a goal of IRIXCE, but we need some FOSS components and specifications filled, which is why I have things like libxg, irixce etc. working to help fill in the "holes" that a modern IRIX distribution needs. Down the line, the eventual goal is to be able to easily bootstrap a system as we require it and have custom code and scripting to handle installations of third party programs, restore config files, etc.

I could technically **probably** create new images now, but I really want to remove obsolete and dangerous shit to lighten the IRIX installs, remove SGI branding, etc. Customizing things would be grand though. Ian's system images are neat, I've spied them before, but it's still not exactly "clean" because things like the SGI ESP are still there, as well as obsolete components.

Anyways, this went off the rails big time.

Back to the topic: Independent enclave of internet.

[jan-jaap]

many aspects of pre-HTML5 pages are deprecated or even not supported, such as the <blink> [...] tags.

Best deprecation EVER

And let's not forget the Hyves Dancing Banana

[vishnu]

The true skinny behind the origin of the blink tag:

http://www.montulli.org/theoriginofthe%3Cblink%3Etag

[Raion]

The only reason I even mentioned the blink tag was out of example. The way in which HTML5 and its predecessors render pages, form divisions and boxes, not to mention integrate CSS is totally different from each version.

I'm not suggesting a return to things like frames or ugly table layouts but there are ways of getting around all of that using clever programming tricks. It's not that much more difficult to program HTML in legacy formats.

[Mario9501]

Man I would love for this to come to fruition, I have some Win9x machines I would love to have on the internet but can't due to security requirements for the web browser that it dosen't have.

[defaultrouteuk]

I’m a bit late to this party.

Having read the comments my takeaway is that you’re essentially describing an Intranet. I also read one comment from @commodorejohn related to the old webring which I also remember. The webring members signed up to a code of conduct (stamp of quality of you will) allowing them to have a logo on their website and participate in a shared directory.

As an old IRIX admin on one of the possibly few intranets hosted on Netscape Enterprise Server (EDIT: on IRIX, in the UK) I have deployed similar closed communities in the mid 90s. The premise being (as you know). - login, create your site, deploy, add a link. All pretty simple stuff. Obviously there are security concerns using that server today but the concept may still be sound. We could offer a wildcard SSL (EDIT: newer browsers) and host under an umbrella domain of your choosing. Users of the community can host under a common framework (possibly remote nodes connected with VPNs back to the central service(s)?) and hosted inside a walled garden approach. It could have the look and feel of an intranet but crucially (and as to the main meaning behind the effort) be available and searchable by the internet.

EDIT2: also I saw a recent video by Luke Smith who had deployed a self-hosted search engine called Searx ( https://searx.me ). I'm not sure how that might fit in but you can restrict what sources are searched etc ... might slot in somewhere to your vision?

I’ve spoken with Raion a number of times and can certainly offer infrastructure support to such an endeavour.

HTH. Rich

[soviet]

Could be even better to be able to access all this information from IRIX, using the old web browsers.