Before the start of this project I suggest a campaign to convert all already existing hardware to run openwrt to reduce electronic waste ;- )
In the U.S. Free market pressures promote planned obsolescence and electronic waste. The projects removal of support for 4/32 devices does not help. Still, it could be possible to send waste devices to a reputable recycler - no children picking through toxic dumps of routers - no dumping them in the Indian Ocean. It could be incorporated into the project. Sort of like the One-laptop-per-Child project.
@Doppel-D , do you know a reputable recycler? Could hardware be loaded with a safe, stripped version of OpenWrt and donated? Perhaps some instructions for participants to flash the old hardware, print basic instructions, package and ship.
The top download I saw was the newifi D2. This may be attributed to the Breed bootloader that will essentially download/install openwrt with a few clicks. Would not surprise me that it is also the top Padavan download
Some of the top devices have 500+ downloads for the month of July. For this project to work, we would need to hit those kind of numbers.
Thanks. I haven't looked at that for a while. So maybe not so useful.
the one and only solid metric you can summize from download stats is that availability is king.
Somewhat powerful, 8 ports, 1GB ram. Why not talk to some manufacturers and make a router of your own. Just a suggestion.
I don't think this will work in any direction.
What about all chipping in some $$ and paying a pro dev to perfect openwrt on a relatively new powerful router?
I think I'd rather have some promotion program for supported machines. I.e. have some trademarked logo "OpenWRT certified" that manufacturers are only allowed to use if their machine is sufficiently well supported (e.g. espressobin), and in return promote (on the openwrt.org web site) those machines which went through the trouble of being certified.
Lets start with basic certification through user poll for the popular Netgear R7800 ;- )
- works best with 18.06.8
- works best with 19.07.3
- works best with master
- tested all versions, you've to live with problems ;- )
Surely superficial because not taken into account who tested with what options / packages / clients / load and what master is really THE master ;- )
To vote, you have to meet trust_level_2
The r7800 should be fine since 18.06.x, I just wouldn't recommend 17.01.x (aside from lapsed security support, of course), because of the incorrect pre-cal handling for ath10k before 18.06.x. Obviously the current version(s) (master or 19.07.x) should be preferred.
I'm quite hopeful for ipq807x to follow in its foot steps.
Main problem of r7800 is WiFi ... About network handle now that we are enabling nss core it's actually good...
Well, that's complaining on a high level - compared to mwlwifi or mt76. Especially QCA9984 is working rather nicely for me (with both ath10k and ath10-ct; qca4019 and qca9886 aren't bad either).
You are right... Mwlwifi was good but now... Don't know about mt76
Ath10k is problematic only with some device.
I would not be in favor of an OpenWrt 'Branded' router for many of the reasons stated above.
I'll add a few more practical considerations...
Bringing consumer products of this nature to market is actually very complicated. It's not just manufacturing (although that is part of it).
- From the engineering perspective, it is surprisingly complex to design a system like this. It certainly can be done, but you need engineers who are experts in several domains. RF alone can be quite tricky and you'd need good RF engineering to build a router that would be competitive in the market
- Regulatory compliance is multi-faceted and in almost all regions of the world, there are specific rules about a product must and must not do. These can be difficult to understand, complex to test, and may be expensive to certify (both official lab testing and filing costs).
- Manufacturing and distribution is not trivial. There is so much that goes into the process including supply chain, manufacturing flow, testing and quality control, and distribution channels are among the many things that need to be managed.
- Customer support, hardware warranty/RMA/replacement all need to be handled.
I've just named a few items... I am an engineer and I work for a company that produces consumer hardware products -- we have big teams working on all aspects of commercialization. Can this be done by a small team (ala crowdfunding/startup projects) -- yes, of course. But it is really hard! And without a range of experts on the team and the resources to produce at scale, it can be almost impossible to release a product on time, on budget, and for the right price.
Not to mention that there are lots of companies that already produce fantastic hardware that is (or can be made) compatible with OpenWrt.
While there is some of this, for sure, it is important to realize that technology marches on. The "free market" element is really about the fact that consumers want newer/faster/better technology. It's not any different than fashion in that sense. Most products are not actually manufactured with "planned obsolescence" in mind -- this implies that by design, the product will fail or become entirely non-functional/useless in a some relatively short time. Some components like batteries are limited by the chemistry, not intentionally designed to fail. Companies must often make choices about support horizons, but that doesn't mean that the device will cease to function (my officially obsolete 11 year old Mac Pro is still my daily driver).
Technology marches on... again. Many (maybe all) of the 4/32 devices are 802.11n or earlier (a good number may even be 802.11g) and have had their lives significantly extended by being able to run OpenWrt.
If you want to run Windows 95 or Mac System 7, 4MB RAM is fine. But with advancements in capabilities, you need more powerful systems with more resources (Windows 10 requires 2GB and Mac OS Catalina requires 4GB).
At some point you have to drop support for the really old stuff so that you can do new things. Good luck trying to watch a basic HD video on a computer that has 4MB RAM.
As far as OpenWrt is concerned, the criteria to avoid planned obsolescence is actually rather simple:
- take a SOC/ wireless driver with okay'ish mainline support (ath9k/ ath10k/ ath11k or mt76 win, mwlwifi fails)
- don't skimp too much on the specifications (flash and RAM in particular) to cut costs (even 15 years ago 8/32 devices were readily available (e.g. Asus WL-500gP v1), around 8 years ago the RAM sizes slowly started to grow in the consumer market)
- a half decent recovery mechanism, to guard against bricking.
- ideally at least the SOC is popular among OpenWrt users and developers (so again, don't fish at the bottom of the barrel, pick a device that is interesting and useful to developers who have their benches full of old stuff)
Neither of these need the additional burden of stamping it as officially blessed (very much on the contrary, for all the reasons that have been raised).
It highly depends on what this project is aiming for:
1.) Whole supply chain (self-made/crafted everything e. g. including wifi-chips). Impossible
2.) Buying components and assemble them to an end-product. Very hard and expensive. But doable if you look at raspberrypi project.
3.) Buying a finished product from a big/solid manufacturer, label it by your own and push your own firmware on the devices. Possible. A lot of companies doing this nowdays.
4.) Buying a finished product. Push your own firmware on it and sell it. Possible.
The last option you can see here:
Have a look at the prices what they are charging. The flashed routers are DD-WRT based and comming along with an extra "App" to monitoring/managing the devices. I don't know how their calculation is looking as they are offering VPN in addition. So it might be that selling those routers only is not enough to keep the business running. But those calculations have to be done before this project is changing over in production.
This didn’t seem to be the model that the OP was suggesting. It seemed like this would be a hardware product designed from the ground up.
If the OP was suggesting a simple white-label arrangement based on an existing producer, how would that make It any different than the user flashing openwrt themselves, except for the fact that it would be preinstalled and badged?
I don't know what will be the end of this discussion/think tank/brainstorming process. But I think the last one is a viable/valid because it would be a starting point. In terms of getting experience about the market, the technic, getting contacts, etc. The higher you enter the higher the costs, the higher the knowledge you need to have/buy, etc. In my opinion you cannot simply enter a market with a new product and hope the best. It is not like that there is no competition like the raspberrypi had on entering the market. That was a new product in many terms. Routers you can buy like bred.
And there are differencies. The customer has serveral certainties/adavantages:
- No knowledge about flashing process at all.
- No risk losing the unit on flash process.
- Customer is not losing warranty/support.
- An assured lifetime in terms of software fixes/updates.
- The certainty that OpenWrt is running. Not like buying and testing if its running.
- Their beloved OS is already on their unit.
Maybe that I've overseen some points.
If a custom made device is to be created from scratch there is a relevant service from 8devices. The fact that they already use OpenWrt makes it easier.