On Sat, Mar 3, 2018 at 11:28 PM, Khem Raj <[email protected]> wrote: > On Sat, Mar 3, 2018 at 2:45 PM, Otavio Salvador > <[email protected]> wrote: >> On Sat, Mar 3, 2018 at 4:59 PM, Khem Raj <[email protected]> wrote: >>> On Sat, Mar 3, 2018 at 9:44 AM, Matt Madison <[email protected]> wrote: >>>> Since it looks like I'll have to do a v5 on my series anyway, I can >>>> pull in that backport for the aarch64 issue. >>> >>> sure, I would suggest we keep 1.9 around as well. >> >> I am against. Go is good to provide backward compatibility and keeping >> both just causes people to delay the update. We have some time to >> April release and thus any regression will be ironed out. Possibly >> wait until 1.10.1 is out but keeping two releases I am unsure it is >> needed. > > Your disagreement is acknowledged. > There are larger systems written in go which haven't moved to go 1.10 > and I know for sure influxdb and grafana are not working with 1.10 > and there might be many more real world programs which will be in same > boat. This is no different then what we do with gcc releases where we > overlap them for couple of releases to smooth out transition.
I understand the need and I see the value. The GCC is harder to upgrade but problems using newer Go toolchain does not happen very often and likely are caused by a regression on Go itself. As I said, there are two options: a) wait for 1.10.1 release b) keep 1.9.4 around My preference goes to (a) as I prefer to not include 1.10 now with those known issues. -- Otavio Salvador O.S. Systems http://www.ossystems.com.br http://code.ossystems.com.br Mobile: +55 (53) 9981-7854 Mobile: +1 (347) 903-9750 -- _______________________________________________ Openembedded-core mailing list [email protected] http://lists.openembedded.org/mailman/listinfo/openembedded-core
