Be warned, there are a number of oddly angry folks here - most of whom don’t own Glowforges
Ha. Not surprised. It’s like the internet is created for oddly angry folks. I feel my frustration is warranted, and I am indeed an owner. With some simple fixes, hopefully not soon a recovering Glowforge owner.
They know people want offline functionality. They don’t intend to offer offline functionality.
Yeah, I completely feel that. It’s an old argument. However, what I’m suggesting wouldn’t be offline functionality. It’s just a mirror on a nearby RPi or whatnot. All it needs is a simple conversion script, not all the fancy features. Steps:
buy RPi.
“nmp install makeGlowforgeSuckLess”
I run a studio from a warehouse space. The internet isn’t great. Eventually, we’ll get fiber. But in the meantime this is what people here say. I’m trying to present as literal quotes:
“The GlowForge isn’t working. Is someone uploading a big file?”
“It’s gonna be faster for me to bike home and use the laser cutter in the maker space over there then bike back.”
“I’ll just cut the piece over at Xs place in the morning then bring it in that afternoon.”
It’s not a hard solution. As a community we shouldn’t accept such a massive design flaw when there are simple solutions.
I thought I heard that GF had released the code that runs the units to the coding community so they can work on GF working off line in certain circumstances … did this go anywhere?
Those people look great. I’ve been watching them. It’s more ambitious than hosting a simple mirror, though.
The thing is, if you do any research before buying, you know all this going in.
Very, very true. I made a probabilistic decision. I figured you get computer-vision-based features. Tracing? Nice! You also get some super weird decision about online insistence from a company created by computer scientists with no hardware background, but how bad could it be? My probabilistic decision did not work out well. But I do have only myself to blame for not thinking of worst-case scenario as assuming that. (Doesn’t help that all the computer vision is an arbitrary amount to the front and left on my machine, so those features are useless to me.)
I feel like this is one of the stages of glowforge ownership. Everyone has this thought at some point, and they are all the lord’s righteous flame about it – for like 2 weeks. I’m not saying your frustration isn’t valid, I’m just saying that it’ll probably go away, like most of us.
The key is to get a good router connection. Once you get a decent router setup and have solid internet and the machine updates automatically, most of us forget all about wanting to mess with the admin aspect, and just get to making.
For me, it was setting up my old N router much closer to the glowforge. Until I caved and said “OK fine, time to dig out that old router and literally mount it on the outside of my house so it can reach my workshop”… I was seeing all sorts of issues. Once my wifi signal was bulletproof, my concerns about the constant on thing went away.
I still haven’t forgiven them for not having an ethernet port, though. That just seemed like a bad call. Yet here we are, and I never worry about the internet thing even in passing – TP-link for the win, amirite?
All that being said, you seem like a sharp person, you might follow @dierdrebeth’s link to openglow, It was a bunch of dudes who said “how hard can this be??” and started their own firmware project. 3 years later, and it’s actually pretty hard, turns out… but head over there and check it out if you haven’t, you might like it. Maybe you’re the missing piece to their skillset that can get it done?
I feel like this is one of the stages of glowforge ownership.
I genuinely wish this was the first time. Or even the second. Multiple helpers at the studio consistently curse at the thing. I get to pay the bill for them figuring it out when it won’t calibrate.
I have the best internet connection available. It’s not great, but there’s nothing I can do about it. That said, the Glowforge is the only internet-requiring gear that has such periodic failure.
There’s a Shapeoko across the studio. The contrast in usability is extreme. The Shapeoko just runs. No fighting. At this point, we’re all happy when the Glowforge finishes calibrating and cloud-processing simple cut files in under two minutes. Being happy at that is essentially Stockholm Syndrome.
I was hoping Glowforge would actually respond to this. Instead, they just bumped the thread to “Beyond the Manual” aka “not our problem” aka “don’t care.” They’ve got my money, I guess.
I’ve even analyzed the JavaScript to see if there are errors on their end just so I could finish a job. Sometimes it takes sooo long even with good internet connection. We’ll see their server spew unhandled JavaScript exceptions. (After talking with them, they said this is normal. I’m uneasy trusting engineering teams with that philosophy.)
Openglow is great. But very ambitious and not a guarantee against Glowforge bricking the device. I have no interest in increasing the usability of Glowforge on my own time and dime. I just want a simple RPi mirror or something to more quickly and reliably this thing.
At this point I just need to sell. I ran the numbers on the time wasted. And the psychological friction of hating the machine and using the Shapeoko instead even if the kerf is a bit too large. I’m out.
If anyone can conscientiously recommend this machine to others, I’ve got one available at considerably less than retail.
You have to have a good Internet connection to use a glow Forge. Simple as that. If you don’t believe you’ll ever have a good Internet connection, then you’re right to get rid of it.
You seem pretty technically savvy. Of course there are two parts here, your internet connection and your wifi connection. If your internet is slow and unreliable yeah you’re kind of stuck. If the WiFi is bad, though…
Anyway I assume you’ve worked on this plenty.
Out of curiosity what speed is your internet connection when you do a speed test on your phone on WiFi when standing right next to the GF? And what about when you stand near the router?
I’m just curious because I’m not sure I’ve seen discussions about how fast is fast enough. Personally I get about 30 megabits down and 20 up, and it works consistently. Fiber would probably be sweet in your space but you generally don’t need fiber speeds for this.
On a stable but consistent, throttled to 128kb connection, I would have slow as hell uploads, but it would have no problem calibrating. Stability is key. Luckily, most of my uploads to the GF were in the 60-70kb range.
Re: 128kb connection, that’s what happens to a Verizon post-paid hotspot plan when the “unlimited” runs out… not deprioritization (which you can get on a pre-paid plan) but straight up throttling down to nothing.
In my experience the idea all devices connect across the Internet with the same quality is a myth. After developing a new device with some non-critical Internet connected features a few years ago we rudely discovered it works great for most of our customers. For the ones with connectivity issues, well there is very little we have found that we can do. They have a quality internet provider and connection, but the connection just doesn’t work all the time. Our device contains a high quality, OTS piece of hardware running the latest Android code. The HAL layer was written by nVIDIA. IMO the Internet at the lowest levels isn’t as great as its hype. Maybe that is an argument for not relying on the Internet, but not relying on the Internet because sometimes it doesn’t work has its own internal logic flaws.
In the 1990s I worked for a data information service where modem banks were the means of connection. Hayes compatible modem problems were the headaches of our life. Modems made by Hayes had no problems. Hayes went out of business before dial-up modems started to disappear. I don’t know the details why, but I suspect it was because a 56K Hayes compatible modem was $50 and a 33K Hayes modem was a couple hundred. For those who remember troubleshooting modem connections the Internet is clearly way better, but I am not convinced it is Hayes better.
Sorry to rant on your post. You just want a glowforge that just works like most of the rest of us have and not my theories on we need a more reliable Internet.
Yes, more reliable internet would be amazing. But even when it’s down to 2 Mbps, the Glowforge can still act up. It’s not possible to tell if its the connection, their servers, or anything else.
It’s worth mentioning again that there’s a quite easy solution. All GlowForge needs to do is have an easily installable server that can run locally. Aside from the internet connection, think of this: GlowForge will not be around forever. Your multi-thousand-dollar investments will brick. It’s not “if.” It’s “when.” A simple local server is very feasible and extremely cheap to drop onto an RPi or whatever.
That solution solves both intermittent connectivity and the fact that eventually we’ll all own fancy paperweights.
Probably because they’ve said in the past that it’s not on their radar at the moment. Multiple times. Like I said to someone else, you knew what you were getting when you bought it, or you didn’t research enough when you bought it.
As for running Glowforge’s server software on a Raspberry Pi, it’s not nearly as easy as you make it sound. I can probably think of a half a dozen complexities off the top of my head…
Most people are not qualified to install and maintain a linux server in their house, who’s going to provide support for that? You think Problems and Support is busy now…
This assumes the local hardware is even powerful enough. GF right now uses Google Cloud servers, we don’t know how much horsepower they’ve provisioned to do all the image processing etc. Do you realize a lot of customers don’t even have computers and run their Glowforge off of a crusty old tablet?
How much time and money would it cost to port the software to run in another environment?
What’s the story for updates?
What about intellectual property concerns?
Are all the code libraries they’re using licensed to be distributed to end users?
Most network problems are WiFi problems, this doesn’t go away with a local server
How does it impact troubleshooting when support can’t look at logs in the cloud?
These are the kinds of concerns that would come up in any product decision for a company the size of Glowforge. There is no “just” do when it comes to this stuff.
My Rasp Pi-3B can barely handle Inkscape and isn’t terribly responsive when used with the GFUI. A couple projects ago we used a dedicated DSP for image processing that drew a full watt of power, I have problems believing a low grade, general purpose, what is it an ARM in a pi?, CPU could handle it.
I disagree, at least in part. It’s sympomatic of the Glowforge community just accepting things as are. There’s that metaphor of the elephant who was chained, stopped trying, then could be tied down with a rope.
I think a simple GCode-to-actual-motor-control-code is quite feasible. Yeah, do away with the tracing or and all that. (Mine never worked anyways.) Just make an executable for an RPi or whatnot. It’s much simpler.
not qualified to install and maintain a linux server in their house,
Let’s not underestimate people. You just have to order an RPi or something akin. Then click the executable.
Most network problems are WiFi problems, this doesn’t go away with a local server
It largely does solve it. It’s WiFi intranet connectivity needing to internet connection 6 inches away. How would it not?
What about intellectual property concerns?
If Glowforge is protective of simple conversion protocols, they shouldn’t be trusted to keep our machines from becoming bricks.
Do you realize a lot of customers don’t even have computers and run their Glowforge off of a crusty old tablet?
Do you realize an RPi is $35?
How much time and money would it cost to port the software to run in another environment?
Very little for a simple conversion script. Will the conversion logic, I could do it in a few hours. DLL or .so if IP for simple conversion is REALLY that important to them.
It does seem this community is very locked into accepting the severe limitations of this system. That’s the community’s call. It’s another reason while it’s time for me to trade up.
you knew what you were getting when you bought it, or you didn’t research enough when you bought it.
I recommend doing a few cuts on a machine that simply works. Then next time you use your GlowForge, use a stopwatch. Count the difference. Multiple that by every cut you’ll do. Add more for servers hiccups, etc. I am well aware that it’s cloud-based. I used many cloud-based solutions. They can be much, much better. There’s this general notion in the GlowForge community of “well, that’s just how it works now. That’s all it’ll ever be.” It’s limiting.
Wrong person to ask that about. I come from the 3D printer world where time for prints are in the hours and sometimes DAYS. I couldn’t care less about a few seconds here and there.
My list was not intended to be a thorough analysis of the subject, merely an illustration that only a few moments’ thought are necessary to scratch below the surface of “just slap a Raspberry Pi on it” and uncover numerous issues that would need to be addressed.
Your reply shows that you are unwilling to invest a few moments’ thought.
Unfortunately it seems you can’t go anywhere without finding the same
Wrong person to ask that about. I come from the 3D printer world where time for prints are in the hours and sometimes DAYS. I couldn’t care less about a few seconds here and there.