Hi Martin, > -----Original Message----- > From: Martin Blumenstingl [mailto:[email protected]] > Sent: Mittwoch, 30. September 2020 20:55 > To: Adrian Schmutzler <[email protected]> > Cc: [email protected] > Subject: Re: [RFC PATCH] ath79: remove model name from LED labels > > Hi Adrian, > > On Sat, Sep 26, 2020 at 6:33 PM Adrian Schmutzler > <[email protected]> wrote: > [...] > > Apart from our needs, upstream has deprecated the label property > > entirely and introduced new properties to specify color and function > > properties separately. However, the implementation does not appear to > > be ready and probably won't become ready and/or match our > requirements > > in the foreseeable future. > I recently did an experiment with the color and function properties and it's > working well for me with my F9K1115V2 to make it work I had to come up > with a (simple) patch that parses the "color" property: [0] > > there's no feedback in this mail - my goal is to understand what the concerns > are so I don't "just" start porting some of my boards to the new way that's > recommended upstream only to see this patch then being rejected in > OpenWrt.
My initial idea was to go the same way; have a look at these two comment where I summarized the response from upstream and why I consider that not helpful: https://github.com/openwrt/openwrt/pull/3437#issuecomment-696377807 https://github.com/openwrt/openwrt/pull/3437#issuecomment-699492366 Apart from that, I'm aware that we obviously have the option to _partially_ implement the upstream solution, i.e. use the color ids and just put the last part of the former label into function. However, as discussed in the referenced comments, this would essentially be an abuse of "function" as far as upstream sees it: We should create the corresponding LED_FUNCTION entries instead of putting strings into "function" directly, and my proposals for adding some of them were rejected. Of course, we could just ignore that, and put into function what we like. That would work perfectly, it's just a little more complicated to implement in user-space as your patch shows. Nevertheless, I just hesitate to implement a system against its intention (color/function) when we have an alternate solution that we can use as intended (label). Apart from that, as stated in the referenced comments, your solution will break anyway when kernel starts to use the devicename strings again for internal "devices" like phy0/wlan0/etc. as proposed in the discussion at linux-leds (e.g. wlan0:green:activity instead of green:wlan2g). As I said in the PR, unfortunately all of this is essentially very unsatisfying at the moment. So, what I plan to do with the labels is as much improvement as I can get for now. Best Adrian > > > Thank you! > Martin > > > [0] > https://github.com/xdarklight/openwrt/commit/eddaa9004f1897688a15befb > d37234f9de1a7ae1.patch
openpgp-digital-signature.asc
Description: PGP signature
_______________________________________________ openwrt-devel mailing list [email protected] https://lists.openwrt.org/mailman/listinfo/openwrt-devel
