Re: [PATCH] Add minimal Hexagon target - First in a series of patches - linux-user changes + linux-user/hexagon + skeleton of target/hexagon - Files in target/hexagon/imported are from another project and therefore do not conform to qemu coding standards

[Richard Henderson]

On 11/21/19 7:01 AM, Aleksandar Markovic wrote:
> On Wed, Nov 20, 2019 at 8:49 AM Richard Henderson
> <address@hidden> wrote:
>>
> 
>> How's that?  He has been asked to split the linux-user stuff from the target
>> skeleton stuff.
> 
> ...
> 
>> This argument would make more sense if there were more present here than a
>> skeleton.
> 
> Speaking about anatomy, I am opposed to upstreaming any "skeletons".
> The other month, another community was dead serious wanting to
> upstream code based on "proposal of the draft" (or was it "draft of
> the proposal"), and now we want to upstream "skeletons"??
> 
> And even that "skeleton" can't be regularly built stage by stage, but
> must resort to "enable configure at the end" cheap tricks?
> 
> What happened to QEMU upstream?

You are over reacting.

We're also talking about *2* patches before the "cheap trick" of enable at the
end.  The only alternative to the "cheap trick" is *1* patch, which no one
wants.  There really is a minimum amount of code that is required before
*anything* will compile.

We're not talking about applying anything at this stage.  We're talking about
patch review, and making the pieces small enough to actually review.

I assume that Taylor and the revng guys are coordinating the actual meat of the
port.  I assume those will be presented in reviewable chunks, each of which
will compile.  Only when we see all of those will the whole thing be applied.


r~