On Mon, 23 Sep 2013 12:42:34 +0300 Artem Bityutskiy <[email protected]> said:
> Hi JF, > > there is a lot of confusion in this thread about what are you going to > do and why. > > And frankly, the announcement is guilty in that. You did not carefully > and nicely explain what exactly you do, and why exactly, and which users > would be affected and how exactly. > > I suggest you guys to come up with a new good survey description. Make > sure it answers the above questions. Also, go through this thread, find > out what confused people, and make sure you have a Q&A section in your > new survey, and the section clears the confusion. > > Thanks! > > P.S. What I found out from talking to people, _not_ from the announce: > > 1. The intended change _only_ affects mic and gbs, nothing else. > 2. The change is about stopping producing i586 mic and gbs builds. Other > tools will still have i586 builds. > 3. The change _does not_ mean that mic/gbs won't be able to generate > i586 images. > 4. On the opposite, x86_68 mic/gbs _will_ be able to generate i586 > images. > 5. So this will only limit people in choosing their _build host_, not > the _target Tizen OS_. > > Right? that's what i'm assuming. :) and here i assume that tizen SHOULD be self-hosting. that means a tizen install (on a hypothetical tizen pc laptop or a tizen tablet or phone or ivi system - who cares) SHOULD be able to, with some added pkgs, have gbs, mic, gcc, toolchains etc. installed just like any distro, and then i should be able to use that tizen running device (be it ix86, x86_64 or arm or arm64) to build tizen entirely from nothing to working image (building tizen here means build a running working target image of ix86, x86-64, arm, arm64 or for that matter any other architectures we will end up supporting, if any). if removing support for 32bit means "giving up on qemu 32bit address space bugs" and thus making 32bit hosts unable to run gbs etc. etc. to run a local build to product an arbitrary architecture output, then i think that would be embarrassing that we can't self host. > > On Mon, 2013-09-23 at 17:33 +0800, Jian-feng Ding wrote: > > Carsten, > > > > All the embarrassings you are talking about just does not exist :) > > > > Current Tizen pkg build is self-hosted, both ARM and IA, both 32bit and > > 64bit. > > > > Current problem we are talking about is the developers' working distro, not > > the Tizen runtime:) > > > > thanks > > - jf.ding > > > > > > On Mon, Sep 23, 2013 at 06:23:37PM +0900, Carsten Haitzler wrote: > > > On Mon, 23 Sep 2013 08:28:06 +0300 Leon Anavi <[email protected]> said: > > > > > > > Hi Daniel, Carsten, All, > > > > > > > > On 2013-09-23 03:46, Daniel Juyung Seo wrote: > > > > > On Mon, Sep 23, 2013 at 8:53 AM, Carsten Haitzler > > > > >> imagine if all the engineers and executives etc. at bmw drove fords > > > > >> because > > > > >> "bmw's can't manage to get us to the office without breaking down". > > > > > > > > > > Awesome analogy. > > > > > > > > > > I am also against dropping 32 bit system support for dev tools. > > > > > > > > > > I bet many people I know use 32 bit system yet. > > > > > > > > > > > > > The example with the cars is nice but I don't think it's very > > > > appropriate. In my opinion it is fine to drop 32-bit support although I > > > > still keep 32-bit distributions on some of my older computers. > > > > > > > > Today nobody uses 16-bit OS for development of Tizen and in long term I > > > > don't expect anybody to be using 32-bit. So dropping 32-bit support is > > > > just a matter of time. Almost all of the CPU for personal computers > > > > produced in the last 5 years are with 64-bit instruction sets. The > > > > majority of developers (including me) that still have 32-bit OS on some > > > > computers actually have a compatible 64-bit hardware and just have to > > > > replace the OS and the dev environment with a 64-bit version. > > > > > > 64bit os's come with a major cost in memory footprint for pointers... it's > > > quite measurable. and that doesn't change the fact that: > > > > > > 1. arm is STILL 32bit and will be for a long time still. if tizen can't > > > be built ON an arm platform (which is 32bit) then it can't self-host on > > > arm... which is embarrassing. > > > 2. reality is that on x86 we likely might support 32bit tizen simply due > > > to the memory footprint issues, and there is still x32 which is 32bit > > > address space WITH 64bit instructions for max perf and best footprint. if > > > we can't work on x32 or ix86 we can't self host. embarrassing. > > > 3. you do know that intel mobile chips (for tablets/phones) are not 64bit > > > still (i thought baytrail may support 64bit in at least some configs, but > > > older ones... no 64bit, and baytrail devices as best i know are still not > > > shipping as of today)... :) again - if we can't self host on these... > > > embarrassing. > > > > > > :) > > > > > > > Best regards, > > > > Leon > > > > > > > > > -- > > > Carsten Haitzler (The Rasterman) <[email protected]> > > > _______________________________________________ > > > General mailing list > > > [email protected] > > > https://lists.tizen.org/listinfo/general > > _______________________________________________ > > General mailing list > > [email protected] > > https://lists.tizen.org/listinfo/general > > -- > Best Regards, > Artem Bityutskiy > -- Carsten Haitzler (The Rasterman) <[email protected]> _______________________________________________ General mailing list [email protected] https://lists.tizen.org/listinfo/general
