[OpenWrt-Devel] [PATCH] ar71xx: add support for GL.iNet GL-X1200

[Jeff]

On 4/11/19 5:56 AM, Petr Štetiar wrote:
> Guilin.Wang <guilin.wang at gl-inet.com> [2019-04-11 17:18:12]:
>
> Hi,
>
>> I want to submit it to ar71xx first, and I will submit it to ath79 later.
> tl;dr version:
>
> please don't waste your (and our time) with ar71xx anymore. BTW it was ath79
> first and ar71xx possibly later, but this period is already over. At least for
> me ar71xx is currently in 'fixes only' mode.
>
> Longer version:
>
> Even if you've provided ath79 support first I would simply hesitate to include
> ar71xx support for this device as it has simply non trivial amount of changes
> (the non-written rule it was agreed upon). Your changeset has following numstats:
>
>   12 files changed, 218 insertions(+)
>
> Please take a look at the recent commits to ar71xx, it's pretty much 'fixes only' mode already:
>
>   93d23aced ar71xx: Correct MAC address for WAN interface of Archer C7 v5
>    1 file changed, 4 insertions(+)
[...]

Wow, I would have thought a simple

"Thank you for your submission. We appreciate your commercial concerns. 
We look forward to your future submissions on the ath79 platform."

would have been more than enough.

GL.iNet is one of the few OEMs that commercially benefit from OpenWrt 
development that take the time to try to return patches and enhancements 
directly to the project. "Don't waste our time" comes through loud and 
clear. If I were an OEM, why should I bother submitting patches? There 
is real opportunity cost associated with doing so, and not very much 
tangible benefit compared to maintaining their own "private" branch.

That the ath79 target on Linux 4.14 doesn't support NAND makes it 
challenging for a manufacturer to simply switch over their entire code 
development. Yes, patches for ath79 on Linux 4.19 dropped a week or two 
ago. However, a reputable manufacturer isn't going to ship product on an 
untested code line. So far I have seen that batman-adv won't even 
compile under ath79/4.19. Also, while the framework is supposedly 
present for NAND in Linux 4.19, as far as I know no devices have been 
demonstrated to be able to use it under OpenWrt. Not that I expect those 
things to magically happen, but they do make it challenging for a 
responsible OEM to switch over as easily as a hobbyist.


Jeff



_______________________________________________
openwrt-devel mailing list
openwrt-devel at lists.openwrt.org
https://lists.openwrt.org/mailman/listinfo/openwrt-devel