AC5300 - With 3.0.0.4.380_3000+ - Merlin Builds Working yet?

[Curado Kev]

Long time reader and ASUS router freak - So I got a new AC5300 after running AC68U for 3 years on Merlin. I have been reading these forums consistently to find out when I can use Merlin, but was on vacation for 2 weeks. I was extremely dissappointed that this occurred after I bought the AC5300.

What is the status now ? If it has not changed and Merlin builds no longer work.

Can I flash the ASUS factory firmware back to pre-3.0.0.4.380_3000 then flash Merlin ? If that is possible, where can I get the "older files" to down grade the firmware ?

Any idea on timing for either ASUS fix or Merlin Build that will be ready for this ? I realize you would post it but looking for the most up-to-date information and thank you for all you do - Merlin !

Thanks for your help !

 

[L&LD]

Try the latest 380.61 Alpha previews of RMerlin's firmware.

When they're ready for your model, of course.

https://www.mediafire.com/folder/bj94sbhrh7e49/Test Builds

Otherwise, you may want to test the slightly older 380.60 Beta 2 builds.

http://www.snbforums.com/threads/beta-asuswrt-merlin-380-60-beta-is-now-available.33199/

 

[RMerlin]

380.60 was supposed to be the first build supporting the new firmware format, however the GPL code on which that version was based had a few WAN-related issue that I can't track down, so the release never left beta. Beta builds are available for all models except for the RT-N66U and RT-AC66U.

380.61 development has started based on a newer GPL version, however at this time the only models that can be compiled are the RT-N66U, RT-AC66U and RT-AC68U. No ETA for a final release or for support for other models, as it will depends on when Asus will make the necessary components available for these other models.

I've been talking with Asus trying to find a way to make things easier for me. As they move an increasing amount of code to these closed source components, it makes it increasingly harder for the closed source components of one GPL version to be compatible with a different GPL release. And due to how their own development cycle works, they very rarely release firmwares for all models at the same time, and providing with the compatible components for all models at the same time can't easily be done either. That leaves me stuck with the problem where I can almost never support all models at the same time anymore, and mixing older and newer code together is also becoming next to impossible.

I'm hoping we can eventually come to some way to make it easier, but for now that's not the case. So at this stage, I'm unable to give any ETA of any kind, and I can never predict either which router model I'm able to support with a specific release.

Just to make sure it's clear: the current issues have nothing to do with the new firmware certification. It's purely caused by the incompatibilities of the increasingly numerous closed source components between the different GPL releases on which Asus is basing their scattered firmware releases. It's a problem that has been increasingly problematic over the last year, and will most likely keep on getting worse. So barring any unexpected development, I will have to rethink my own development cycles. So far, I haven't come up with any satisfying plan either.

 

[L&LD]

RMerlin, thank you for that insight.

In the end, you can only do what you can. Whatever that is will still be appreciated by all.

Otherwise, we (now) better see how the future will look. Even for Asus routers.

I really didn't think it would be simply a matter of 'not being able to flash earlier firmware' as the only issue the more restrictive rules would have implied for us users. And I'm sure this fragmentation of your firmware releases won't be the last of the consequences because of those changes either.

Hope that Asus gives you a direct and inside track to these closed source components, as needed.

 

[Curado Kev]

Thanks to all for their insight - Merlin that helps us understand it and thanks for all your time on writing that reply. I usually check in here a couple times a week so I'll keep monitoring it. It sucks I spend $400 on router and can't get your firmware, hopefully that will change soon and ASUS will realize that your firmware is selling advantage for their products. )

 

[Marc66]

I'm hoping we can eventually come to some way to make it easier, but for now that's not the case. So at this stage, I'm unable to give any ETA of any kind, and I can never predict either which router model I'm able to support with a specific release.

Maybe the solution is to have multiple parallel versions; each not necessarily supporting all devices.

If I look at the recent builds, 380.60 supported some models but left out some. 380.61 supports some of the left out models but not others. You may end up with something like:
- 380.60 supports X,Y,Z models
- 380.61 supports A, B, C models
- 380.62 supports A, B, Y, Z models

In my case, 380.60 was not available for my AC66U but 380.61 is. So My firmware path would be 378.58, 378.59, 378.61. For someone else, it would be 378.58, 378.59, 378.60

Of course, you don't want to maintain multiple code bases, but maybe the solution is to have 2; 3 at most. Logically, Asus should release firmware versions in a similar way. To make it easier on their own development cycle, I would expect they have 3 versions: one for MIPS (66U, ...), one for older ARM (68U, 87U), one for newer ARM (88U, 53U, 3100).

You already "sort of" took that path already. Just need to take 380.60 out of beta > That'd be one final release. And keep 380.61 moving, also to a final release. Until then, you may have started to work on 380.62, but at least you have final releases that are close to Asus' latest versions and you support most devices. I would personally be very fine with such an approach. I don't mind running a final version that is a few months old, but it has to be Merlin, not Asus.

Something else you may want to start doing is to compile a matrix with Models VS supported Merlin versions. I wouldn't mind doing this but I would have to search through change logs to make sure I don't miss any model.

 

[RMerlin]

Maybe the solution is to have multiple parallel versions; each not necessarily supporting all devices.

I considered it, and it would be a nightmare to maintain. For starter, I can't rely just on router families. The RT-AC56U for example isn't always updated at the same time that they update the RT-AC68U, despite the two having very similar architectures. This would require one separate branch for each model.

I have a general idea on how their code is "branched out" at the moment internally, and that wouldn't work for me.

The best I can do is just release for whatever model can be supported at specific points in development. It still makes things very hard for me, as I don't want to switch my main router every few weeks to ensure I can fully test whichever model is actually being compilable at the time.

Developing on trunk and branching out every release would make the whole codebase a real mess, in addition to requiring me to merge changes in multiple branches - I dropped the idea. The existing development is already getting out of hands for a single developer to deal with, I cannot afford to make it even more complex.

Just need to take 380.60 out of beta

380.60 will not get out of beta due to WAN issues that I'm unable to track down or debug. I don't issue non-beta releases with bugs I consider to be hard blockers.

If things keep evolving the way they are, it might eventually make this project virtually impossible to maintain in any sane way by a single developer. This could lead to a situation similar to DD-WRT, where one single developer does all he can do to keep developing, and someone else has to take care of making actual releases out of that code. And that's a scenario I'm not comfortable with either, unless I had a REALLY dedicated person handling the release process. Which at this point seems highly unlikely.

 

[Marc66]

I considered it, and it would be a nightmare to maintain.

Understood, ... it was just an idea [emoji53]


Sent using Tapatalk