On Wed, Feb 09, 2022 at 06:19:01PM -0600, Rob Herring wrote: > On Wed, Feb 9, 2022 at 10:04 AM Simon Glass <sjg@xxxxxxxxxxxx> wrote: > > > > +Rob Herring > > > > Hi David and Rob, > > > > On Tue, 8 Feb 2022 at 23:31, David Gibson <david@xxxxxxxxxxxxxxxxxxxxx> wrote: > > > > > > On Tue, Feb 08, 2022 at 02:43:44PM -0700, Simon Glass wrote: > > > > Hi David, > > > > > > > > On Sun, 6 Feb 2022 at 21:10, David Gibson <david@xxxxxxxxxxxxxxxxxxxxx> wrote: > > > > > > > > > > On Sun, Nov 07, 2021 at 03:43:42PM -0700, Simon Glass wrote: > > > > > > Note: This was last sent 6 years ago. It really belongs upstream and I > > > > > > believe it is useful functionality for libfdt, so I am trying again. > > > > > > Please take a fresh look at this. It is cut back from the previous series. > > > > > > If accepted we can expand the feature set piece by piece from here. > > > > > > > > > > Sorry it's taken me so long to look at this. Again. I can't dispute > > > > > that it's useful for certain use cases. But as for belonging > > > > > upstream... > > > > > > > > > > This series adds quite a lot of conceptual complexity. It introduces > > > > > a new data structure, new state structures, a entirely new mode of > > > > > working with a tree and a bunch of configuration parameter types on > > > > > top of the new entry points and new tool. I still find the semantics > > > > > of the different criteria for inclusion/exclusion from a region pretty > > > > > bewildering. > > > > > > > > It is sufficient to achieve its purpose, but I don't think it is any > > > > more complex than that. > > > > > > I don't disagree, but that still ends up being quite complex. > > > > > > > > That makes me pretty disinclined to add this to the scope of > > > > > maintenance for libfdt. As you've probably noticed, I'm already > > > > > struggling to keep on top of maintenance for the existing libfdt > > > > > interfaces. AFAICT everything here can be implemented fairly > > > > > naturally in terms of libfdt's existing public interface. so I'm not > > > > > really seeing a compelling reason for this to be merged into libfdt, > > > > > rather than being its own separate library that depends on libfdt. > > > > > > > > Are you suggesting: > > > > > > > > 1. that libfdt should move to a new maintainer > > > > 2. that you would accept these patches if someone else maintained them > > > > within the libfdt tree > > > > 3. that we set up a separate tree to fork libfdt, with these changes in > > > > 4. that we put these changes in a separate tree? > > > > > > Right now (4) is what I'm suggesting. Or to be more precise, creating > > > a new repo with "libfdtrange" or whatever, that depends on libfdt. > > > > > > (1) and/or (2) are potentially worthy of further discussion. (3) is > > > just a bad idea, IMO. > > > > Where are things going with device tree validation in terms of > > source-code location? > > Right now it doesn't use libfdt at all, but that's what I'm working on > if we can get pieces in place to package pylibfdt sorted out. If not, > I may be doing 3 just for packaging. It doesn't seem to me that the validation should be using libfdt, except in a limited capacity for an initial read. Schema checking is likely to do lots of random access to various nodes and properties, so pre-reading the dtb into a "live" tree format seems like it would be a good idea. > > Is it likely there might be a separate tree for > > that, which could perhaps hold other libfdt dependent things? > > It is a separate tree[1], but it's a pure python project and I don't > think it's the right place for a C library. But we can setup a new > project in the GH devicetree.org group[2] if you want. > > Rob > > [1] https://github.com/devicetree-org/dt-schema/ > [2] https://github.com/devicetree-org/ > -- David Gibson | I'll have my music baroque, and my code david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_ | _way_ _around_! http://www.ozlabs.org/~dgibson
Attachment:
signature.asc
Description: PGP signature
