> -----Original Message-----
> From: Tomasz Maciej Nowak [mailto:[email protected]]
> Sent: Montag, 4. Mai 2020 20:45
> To: [email protected]; [email protected]
> Subject: Re: [OpenWrt-Devel] [19.07] [PATCH] ath79: add support for
> Sitecom WLR-7100
> 
> W dniu 04.05.2020 o 20:31, [email protected] pisze:
> > Hi,
> >
> >> -----Original Message-----
> >> From: openwrt-devel [mailto:openwrt-devel-
> [email protected]]
> >> On Behalf Of Tomasz Maciej Nowak
> >> Sent: Montag, 4. Mai 2020 19:49
> >> To: [email protected]
> >> Subject: [OpenWrt-Devel] [19.07] [PATCH] ath79: add support for
> >> Sitecom
> >> WLR-7100
> >>
> >> Backported from: 1bc921f419df508c57dc07cd3e43cdf0408c17dd
> >
> > Device support is typically not backported to stable branches, unless you
> have a very good reason for it.
> 
> I see bunch of devices added during rc phase and after stable release, so
> what's the truth here?

Okay, so you want the long answer:

In general, stable branches are meant to only receive "fixes", but not 
"features".
In that terminology, new devices are also considered "features", of course.

Backporting a device from master (or reviewing it) frequently needs a 
substantial amount of work and care, as it is linked to several different parts 
of the code which all might have changed in the meantime. That's why 
"typically", device support is not backported. This saves a lot of review time 
that can be spent on other stuff that will go into master (despite, if we'd 
backport devices to stable on a regular basis, it would defeat the idea of 
having stable branches).

However, as you have found out, there are exceptions to this "typical 
behavior". In a nutshell, an exception is made whenever a committer decides to 
make one, and invests his/her time to review or create a device backport. 
Personally, I tend to make exceptions in the following cases:

1. A particular device is just a stupid clone of another device, and 
backporting would complete the family in the stable branch (i.e. add WDR4310 to 
ath79 where WDR4300 and WDR3600 are already present). Reason: Quick/easy 
(because clone), benefit (family complete)

2. I have backported a device for downstream anyway. (If I've done all the work 
anyway, there is no reason to not merge it so others can benefit as well.) 
Reason: No additional work.
Note that this is different from merging the device support somebody else 
provided: In that case, I would need to review it first, and reviewing a 
backport is about the same amount of work as needed for creating it in the 
first place.

3. There is a strong reason of any kind why a particular device should be made 
available. (That would also make me review a more substantial support backport 
from somebody else.)

But as Daniel stated, backporting device support is not "forbidden". That's why 
I wrote "is typically not backported", as it's effectively a matter of 
perceived behavior in the past, and no formal rule.

After all, you just have to find a committer who reviews and applies it. I just 
tried to express that this most probably won't happen, as other committers 
might follow a similar reasoning as discussed above (as experience tells).

So, please don't feel pestered.

Best

Adrian



> 
> >
> > Best
> >
> > Adrian
> >
> 
> --
> TMN

Attachment: openpgp-digital-signature.asc
Description: PGP signature

_______________________________________________
openwrt-devel mailing list
[email protected]
https://lists.openwrt.org/mailman/listinfo/openwrt-devel

Reply via email to