Perhaps it would help if the core developers give some more guidance/insight into planning. For example, namiltd and graysky2 have been doing a lot of work in the past weeks to prep kernel 6.12. They're asking for reviews, and keep rebasing their commits, but nothing is happening. If anyone would say (for example) "thanks for the effort, but we won't be looking into this until after the release of 24.10" then community members know what's up and how to better support the project with their time and effort.
2 Likes
In that case, think even I replied that until 24.10 is at least branched 6.12 is not an option.
For 6.12 its not ideal time currently, as moving to it makes backporting stuff to 24.10 much harder, so I dont think we will merge it until at least final 24.10.
I will probably post a question on the list to see what is the general opinion but please understand that reaching a consensus on a move takes time and we all have limited time.
Previous month or so was basically getting 24.10 ready and fixing up APK migration.
6 Likes
Totally understand, and very valid points. It's exactly this kind of feedback that would be welcome on pull requests 
3 Likes
This is completely understandable, however before you wrote this, I honestly wasn't sure that this was the reason. And I assume that many of the contributors don't know that either, and explain lack of response in other ways, such as: my contribution is unwelcome here. So even a bot response is better than nothing. IMO.
That said, if there is such a large disparity between required and available manpower, then perhaps the situation could be improved by inviting more people for certain roles in the project? There is a fairly large community out there, and I assume that some people would be willing and able to help.
Edit: how about having a list of "help wanted" and when posting the boilerplate response in Github PRs, include a link to that list? This way, rather than discouraging potentially good and capable contributors, you invite them to direct their effort where it is most needed.
3 Likes
I've been using custom firmware on and off for 20 years. I found dd-wrt and Tomato much easier to use than openwrt. Specifically, I think openwrt luci should reach feature and ease parity with major AIO router+AP vendors (linksys).
btw, LucI is a terrible, terrible, horrible name. You could have a gun to my head and I couldn't tell you what it means. Even after glancing at the wiki page I don't know. Regular people, who aren't engineers, like having names for things they use constantly. The lucI vs lucL ambiguity is absolutely atrocious. Even something as simple as Lucy would be much more memorable and avoid pointless ambiguity. Clearly the name doesn't mean much (i.e., isn't an important abbreviation) if the wiki can't even explain what it is.
I have avoided guest wifi entirely because it's a huge pain and doesn't work in openwrt.
I also bought a dumb AP because wifi in owrt is awful.
Me buying a dumb-ish (non owrt) switch is also possibly related.
CAKE / traffic shaping / QOS (many ways of reaching anti-bufferbloat) for me is the killer feature of owrt. it is extremely powerful to tell someone "you don't need a new $300 router, you need better software on your existing $100 router" (enthusiasts at 1Gbps excluded). owrt could be the de facto bufferbloat, n00b friendly, "smarter way to internet" firmware. instead it's much closer to a bare Arduino breadboard and some spare wires. It's extremely powerful and extremely complicated. This is a huge missed opportunity and cuts openwrt's potential reach down to about 1% of what it could be. There isn't a person I know that I would recommend owrt to. Back in 2005 everyone and their mother was installing dd-wrt.
These things should be default:
- anti-bufferbloat
- 1.1.1.1 DNS or similar privacy-focused DNS. I can't remember if I ever asked why DNS isn't defaulted.
- DNS over HTTPS (DoH) default
- other hardening I'm not aware of
owrt could be "the private and secure router firmware" but it ignores this market and lets people sink in a sea of complication. this has real implications for end users and makes them less safe. Why? Anyone who knows what they're doing can change these defaults. We should protect those who don't know what they're doing.
bonus:
- admin over hardwire toggle in webui. once your system is set up, and hopefully for most people on openwrt, you won't need the web console on wifi
- guest wifi with minimal ability for guests to enter the main network
- one or multiple standard boards (the new dev board is very exciting, kind of ambiguous to end users right now). Every router I buy I get told "oh lol, not a lot of people use that, good luck supporting it".
2 Likes
Obviously you have some very strong opinions about your choices in wifi-router equipment. i do believe a dell wyse 5070 is even more powerful than what you are talking about if configured properly. A 4K video stream takes between 30-35mbps. At 1080p max 8mbps. i usually force all my children's tablet devices to download at 720p (~2mbps) just to keep the data rate down since my home internet is capped at a specific GB/month plan, and if i go over that cap they will automatically charge me more money. I'm not really even sure at this point more then a 100mbps connection is really necessary. I mean i own many 4k tv's (6 or 8) but most are still max 1080p or less (20-25) and my children prefer their tablets anyhow! And i don't really care about the difference between 4k and 720p in most of the content i stream anyways. Also most atsc broadcast content is maxed at 1080p and thats only for the main stream and secondary streams are at 720 or 480.
But anyhow, where i live a wyse5070 can be had for as little as $40 on ebay, a 4port ethernet card another $40, and an asia rf 6G wifi card another $40, so $120 for a working device itself, add more if you want more ram or hard drive space and at that price point there is no comparison!
As argued before, we have no way of doing that in a way that works for everybody automatically. In addition this requires a few policy decisions a network's administrator should make explicitly.
This is a question of your local threat model, you might mistrust your ISP more than cloiudflare or the other way round... again this is a policy decision. (And there are corner cases, like e.g. my ISP who resolves its SIP servers names only on its own DNS servers and only to requests coming from within its own network"s IP ranges, defaulting to 1.1.1.1 would result in telephony not working). Now, I am all for giving people sufficient information to make educated decisions, but some decisions they need to make themselves, no?
Maybe, just maybe, first get a feel what is currently well supported, before buying a new device?
2 Likes
I assume it is a dual word shortening of Lua Control Interface.
But as with all shortening's it must sound nice and not only look good, so I guess we ended up with “Luci” a long way back.
2 Likes
I would like to see OpenWrt supports "ldconfig" and other development functionality that is limited by the use of musl. The routers nowadays have the hardware capability (ram and the speed of muti-core cpu) to do fancy stuff that those ancient single-core 4mb-routers can't. I singled out "ldconfig" from my personal experience and others. This particular document - Building OpenWrt ON OpenWrt - is a hint of the direction that OpenWrt can consider moving toward to.
You mean something like ansible or chef/puppet or whatever all those names are, or from Windows it is DSC (desired state configuration)?
Something to remotely ascertain that the device is in a specific state in terms of packages installed and configuration?
hi everybody !
a simple way to create a mesh between 2 router
I personally want to see a fork specifically for x86 hardware where the usual limitations (memory and space constraints) don't make the system impossible to upgrade, meaning you are stuck with old package versions until you reflash a newer image and start from scratch.
1 Like
My up vote to Mercusys MR80X V3
Should be doable
Is it possible to get a template or default config system into UCI?
Like, I have 16 different access vlans and everything is the same on each dnsmasq dhcp instance. Only the nameserver differs. It would be great if I could save me from repeating 20 settings or so but instead point to a default instance where the settings are inherited.
Use the image builder and bake in the config. DD the disk, reboot, done. 
Use a for loop in bash and uci. Then you can create them easily.
I already have a bunch of wrapper scripts, that's not the issue.
But you are right, I can easily create my configs however I like and don't depend on UCI for it.... Indeed 
Seems it only comes in a light mode. Ow my eyes. Couldn't find a way to enable dark mode in it.
You can fake it, just save this somewhere and run it in /etc/rc.local. Change the brightness and rotation for various Shades of Grey.
#!/bin/sh
# vim: set expandtab softtabstop=4 shiftwidth=4:
#-------------------------------------------------------------------------------
CSS=/www/luci-static/openwrt2020/cascade.css
if [ ! -e $CSS ] ; then
echo "You don't have the OpenWrt2020 theme installed."
echo "Run 'opkg update && opkg install luci-theme-openwrt-2020' and try again."
exit 1
fi
echo -n "Modifying $CSS ... "
if grep -q hue-rotate $CSS ; then
echo 'already has dark mode.'
else
echo -n 'making it dark ... '
cat >> $CSS <<EOF
:root
{
--secondary-bright-color: #bbc;
}
html
{
filter: hue-rotate(180deg) invert(1) !important;
background: black !important;
background-color: black !important;
}
EOF
echo 'done.'
fi
4 Likes
You might want to try Argon, then. It's mobile friendly, has a dark mode, and IMO is the best looking OpenWrt theme I've found. However, it's not available in the OpenWrt repositories, so you have to sideload it from a locally downloaded file. For me, that inconvenience (having to do it again each time you update OpenWrt) was reason enough for me to go back to the OpenWrt 2020 theme.
There's also luci-theme-material, which is in the repo, but I haven't used it, so I'm not sure if it has a dark mode or not. I do remember that it's mobile friendly, and based on Android's Material UI.
2 Likes