Hi,

it is not really clear what exactly you want to archive here. As far as I can 
see everything we want to bootstrap we can bootstrap. Not sure why you want to 
completely mess with the order and steps. Maybe write the code and send a patch.

        René

> On 22. Nov 2019, at 03:49, scsijon <[email protected]> wrote:
> 
> 
> My initial thoughts of doing it this way was to build a package containing 
> this toolchain output to become the (sort of) single package at a new stage 0 
> and build on top of that by moving the stage 0, 1, 2 into stages 1, 2. 
> However after some reflections and a couple of online discussions, I no 
> longer think it's practical for what I'm trying to do with T2's build system.
> 
> To give you an idea of my intent>
> 
> musl/clang/llvm now builds after I manually >> build/install LLVM (Pass 1) -> 
> Clang (Pass 1) -> compiler-rt -> libunwind -> libc++abi -> libc++ -> chroot 
> and again build/install LLVM (Pass 2) -> Clang (Pass 2) -> lld -> lldb -> 
> compiler-rt (Pass 2) -> libunwind (Pass 2) -> libc++abi (Pass 2) -> libc++ 
> (Pass 2) chroot again and again build/install LLVM (Pass3) ->Clang (Pass 3) 
> and then everything else again so it all should be musl/clang/llvm only. Only 
> then I believe are you free (as possible) and ready for T2's stage 5.
> 
> Question for Rene if I may, please;
> 
> Can 'we' (I mean you of course, I don't know your system code good enough to 
> start to work out how to do it) add settings to the builds to also do 
> chroot's at stages 3 and/or 4 somehow, I know it's a pain, but it seems to be 
> the only way to be a totally clean outcome, chroot'ing once at Stage 2 is not 
> going to work, problems appear later with some (multiple) packages. It works 
> manually building this way, or is that what your manual is talking about in 
> the build stages on page 46 (index) 36/37 (page numbers) when talking around 
> stages 3 and 4 and i'm misunderstanding your system (again).
> 
> I admit re-chroot'ing a third time and running a Pass 3 run at this point 
> before building other packages is me being extra cautious, but at this stage, 
> well..... I'm trying to keep the possible problems to a minimum, and 
> musl/clang/llvm is new enough that a lot of packages need help (patches) with 
> sorting things out.
> 
> thanks
> 
> On 22/11/19 09:47, Jan Rovins wrote:
>> Yes,
>> A stand alone tool chain would be a nice thing to have, good to use for 
>> embedded application development if you don’t care about the whole Linux 
>> build system at the moment. It could also be modular, so you can build a 
>> tool chain first, give it to team members for application development, and 
>> then later, when you are building a whole Linux distrio you could say: use 
>> this tool chain that I have previously configured.
>> Just some thoughts.
>> Jan
>> Sent from my iPad
>>> On Nov 21, 2019, at 2:38 PM, scsijon <[email protected]> wrote:
>>> 
>>> 
>>> Just looking at the 00-bootstrap file, I was wondering just how many of the 
>>> 57 packages in it could be stripped out of it if you only wanted to create 
>>> a Toolchain to chroot into and work from that, there seems to be an awfull 
>>> lot of packages in it?
>>> 
>>> Stone and Rocknet of course (sorry rene), ?but what else, before adding 
>>> others you may need for your own build.
>>> 
>>> Any thoughts T2and's.
>>> 
>>> 
>>> 
>>> ----------------------------------------------------------- If you wish to 
>>> unsubscribe from this mailing, send mail to
>>> [email protected] with a subject of: unsubscribe t2
> 
> 
> 
> ----------------------------------------------------------- If you wish to 
> unsubscribe from this mailing, send mail to
> [email protected] with a subject of: unsubscribe t2

-- 
 ExactCODE GmbH, Lietzenburger Str. 42, DE-10789 Berlin, https://exactcode.com
 https://exactscan.com | https://ocrkit.com | https://t2sde.org | 
https://rene.rebe.de

----------------------------------------------------------- 
If you wish to unsubscribe from this mailing, send mail to
[email protected] with a subject of: unsubscribe t2

Reply via email to