I appreciate the thought, but I have to disagree there. For every single fix that I do, Asus fixes 10 more issues on their end, while also adding 1-2 new features, and all of this while supporting about 20 different models that spreads over two completely different hardware platforms (I only have to support 6, and they're all Broadcom-based).
There are large portions of the code I refuse to touch simply because it's too complex for me to fully understand. The Dual WAN stuff for example.
Frankly, a big part of the problem there is the Tomato base on which Asus has built Asuswrt. Tomato design is poorly designed when it comes to adding new features since the code isn't modular at all. Asus had to add tons of conditional code on the Tomato base to be able to add new features and support different models (all with different feature sets).
If it were my call, I'd say Asuswrt should be rewritten nearly from the ground up, based on a more modular approach, and with a sane build environment rather than the mess Tomato's build enviro is. Tomato's monolithic architecture frankly blows when you start having more services added on top of it. You end up with services such as dnsmasq being re-started 3-4 times during the initial boot sequence, for example. The firewall rule management is also all over the place, with rules done by the firewall code, other sections of the code insert their own rules here and there, while some section will flat your restart the whole firewall subsystem so it can add the specific rules required by that section.
Overall, the problem isn't Asus's developers lack of skill. Look at the amount of features they have developed compared to any other competitors. Writing some of these features such as the actual Dual WAN system requires some serious skills - there's a reason why such a feature isn't available on any of the open source projects at this point.
