Wifi Extension Development Thread

[hfrazier]

If you haven't already bought a Vonets, then I would recommend an OpenWRT device (see previous reply a bit further back in this thread for more info). The only device I've tested is this one, but it's pretty good:

https://www.gl-inet.com/products/gl-ar300m/

That and it's way more capable and expandable if you know your way around linux!

 

[olePigeon]

I just get so giddy seeing a Wifi menu in System 7.  XD

 

[CC_333]

That and it's way more capable and expandable if you know your way around linux!

It also has the virtue of being a bit cheaper and easier to find

I think I'm going to give this a try!

I'm wondering, though, since this is almost completely platform-agnostic (the only requirement is an Ethernet port), this could theoretically work on PCs.

This is a bit off topic, but one could take this concept and write a comparable program for Windows. Couldn't they?

Having something like this for DOS, Windows 3.x and 9x would be very nice!

c

 

[hfrazier]

This is a bit off topic, but one could take this concept and write a comparable program for Windows. Couldn't they?

Having something like this for DOS, Windows 3.x and 9x would be very nice!

Oh, absolutely! Just a matter of writing the software.

 

[CC_333]

Oh, absolutely! Just a matter of writing the software.

Someone should write it then!

I would, but I'm not that good at programming. Yet.

Aside from the UI bits (which are quite Mac-specific), is your code relatively portable?

EDIT: Just realized that it probably makes heavy use of MacTCP/OT specific things, so probably not.

c

 

[ants]

Aside from the UI bits (which are quite Mac-specific), is your code relatively portable?

Well it's written in C++, so that's a good start. There's a good amount of abstraction in the code too. Probably not something I'd ever tackle though...

 

[Trash80toHP_Mini]

Is there a specific Ethernet card that you need to use for this to work? Also, have you been able to try this with one of the antenna-less Vonets var411n units? I wouldn’t mind having one or two shipped to you if it meant that we could have this thing work seamlessly someday without an antenna sticking out.

I have grave doubts about feasibility of using an internal antenna in a compact. The grayish/silverish paint on the inside of the case is its RFI shielding. Dunno if it's an effective Faraday cage at WiFi frequencies though. It's meant to corral frequencies in the 8MHz to 50MHz range, could it be transparent when it comes to Wifi frequencies?

Has anyone tested the use of a WiFi antenna inside a Compact Mac case? @ants, IIRC you said the Vonets board worked without the antenna hooked up when mounted at the backplane. Was that due to a leaky RFI shielding config or the higher freqs?

 

[CC_333]

@Trash80toHP_Mini Interesting point. When I got some peel 'n stick internal antennas for my Basic Beige Sandy Bridge machine, I was worried that the sheet metal case would interfere with the signals. So far, Wifi has been working perfectly (I'm using it as I type, in fact), so my fears were probably without cause.

c

 

[Crutch]

Curious if anyone has found a router that allows connecting to 5GHz WiFi with the ability to manually configure speed/duplex and disable autonegotiate as is required?  The Vonets device is nifty but 2GHz only (as is the GL-INet OpenWRT AR-300M suggested).  I tried the GL-iNet AR-750, which has dual-band support, but it has two ethernet ports on a virtual switch under eth1, so OpenWRT "ethtool" can't apparently be used to configure manual speed/duplex/autoneg on individual ports (and "swconfig" doesn't seem to let me do it, either).  Next I'm planning to try one of the various OpenWRT-supporting range extenders from Netgear or GL-iNet that only have single ethernet ports, but if anyone's done this already, advice welcome.  Obviously I could set up an additional 2GHz signal but that seems like an inelegant solution ...

 

[ants]

"ethtool" can't apparently be used to configure manual speed/duplex/autoneg on individual ports

I had the same problem with my AR-300M - ethtool wouldn't identify eth1 (the LAN port) - but I later discovered that it worked fine for eth0 (the WAN port).

So then all I did was swap the LAN & WAN ports using the Luci interface, under Network settings.

 

[Crutch]

I had the same problem with my AR-300M - ethtool wouldn't identify eth1 (the LAN port) - but I later discovered that it worked fine for eth0 (the WAN port).

So then all I did was swap the LAN & WAN ports using the Luci interface, under Network settings.

THAKN YOU @ants ! ... this worked like a charm, including with my AR-750 so I now have perfect 5GHz Wifi access from my SE/30 through a OpenWRT router.  It didn't seem to work at first but after I also did this step https://wiki.openwrt.org/doc/recipes/relayclient it's perfect.  Not sure if that should be necessary, I will try fiddling with it some more and see if I can do without the added bridge interface.

 

[ants]

@Crutch awesome! Have you tried out the Wifi extension on your SE/30 yet? I'd love to know if anybody is having success with it.

 

[Crutch]

I tried it this week (thanks for writing it) but oddly am unable to figure out the IP to use for to access the OpenWrt router from the SE/30! SE/30 is on the net and works fine (can even browse m.cnn.com with iCab) and can see my main router at 192.168.1.1 from it, but clueless how to find the IP for the OpenWrt router as seen from the SE/30 even though I can ssh into it from my modern iMac over my main WiFi with no problem. Maybe because I had to set up relayd per the link above, could that be “hiding” the GL-iNet router from the LAN somehow? Bit of a networking noob but this seems weird to me. 

 

[ants]

@Crutch the default gateway IP on the GL-iNet devices is 192.168.8.1, so start by trying that? But if you've bridged your networks, perhaps the device is getting an IP address from your main router.

The GL-iNet also provides a hostname that you can use instead of IP address, on the AR-300M the default hostname is gl-ar300m, so I assume it would be gl-ar750 for your device. Try that if the IP address does not work. I believe the hostname can be viewed in Luci, although I don't have the device with me right now to check.

Let me know if you get it working, and I'll update the Readme in the MacWifi project with this info thanks.

 

[Paralel]

Now someone needs to convince the iCab guys to update their 68k version to the latest release...

 

[techknight]

Now someone needs to convince the iCab guys to update their 68k version to the latest release...

Or open source it. That would be the ultimate solution. 

 

[CC_333]

Indeed.

However, even in its dilapidated state, iCab is significantly more recent than both IE and Netscape (it was last developed during 2005-2008 for 68k Macs, it seems; the PPC version lasted a bit longer).

iCab is still under development, so it's not dead. It would be nice to see the old Classic Mac OS versions (both PPC and 68k) be released as freeware at the very least (as I recall, there were nag screens encouraging people to buy a license, which, aside from IE, wasn't *that* uncommon until at least the early 2000s, but by 2008, it seems most people expected browsers to be freeware).

c

 

[ants]

Even with a hypothetical new version of iCab, in order to make web browsing on a vintage Mac possible you're going to need:

1. An external HTTP to HTTPS proxy - since most of the web is now TLS encrypted with complex ciphers (even a 68040 doesn't stand a chance decrypting a ECDHE cipher before the remote server would time out)
   
2. An external web rendering proxy - since the task of parsing & rendering modern CSS is so CPU intensive
   

Both of these exist, somebody just needs to piece it all together.

And thanks to the advent of the mobile web and responsive web design, if these problems were solved then a mobile-optimized website might display quite well on a vintage mac. The HTML viewport width of an iPhone X is just 375 "pixels" (albeit at a much higher resolution/sharpness, since the actual number of pixels on an iPhone X is 3 times this resolution). The width of an SE/30 screen is 512 pixels (actual pixels in this case). So, ironically in 2018 there's a better chance of web pages working on a vintage mac than 10 years ago!

 

[Paralel]

I think a system running a full 040 might have a chance with at least 32 megs of RAM. Running a recent version of a decent web browser, the Amiga crowd is able to get a modern website to render using the setup mentioned in 20 seconds.

 

[techknight]

Websites these days are horribly RAM hungry, especially with all of, you guessed it, javascript. 

For shits and giggles, a few months back I pulled out an older P166 with 48MB of RAM. I was able to run a relatively modern browser that could render modern websites. Trouble was, it ran out of RAM and kept spinning up the virtual memory, taking FOREVER to load the site. 

I think that was the biggest issue. Even on my G3. I maxed out the RAM on my G3 awhile back and noticed a HUGE improvement in performance with tenfourfox. I have learned if you do not at the bare minimum have 1GB of RAM for the browser to "fill up" when parsing/rendering, your going to have serious issues. 

I dont see a 68K or any machine under 1GB of RAM able to render modern sites, without at least some sort of helper CPU to do the heavy crunching and even to just be used as memory. 

I will probably get flamed for this, but honestly in my opinion modern websites being this way are the result of "bad programming" and dependency-hell. There was a good video I watched on youtube the other day about that very topic, about how we need faster computers to run slower software at the same speed. I think the link came from here even. 

It actually amazes me that computers today with modern software really dont seem any faster than old computers I use with period correct software. Now granted old software systems dont have the same features modern ones do, plus trendy changes (UI), etc.. However the magnitude of the change doesnt warrant the "bloat" of the change. But I digress.