All of lore.kernel.org
 help / color / mirror / Atom feed
From: Bruce Ashfield <bruce.ashfield@gmail.com>
To: meta-virtualization@lists.yoctoproject.org
Subject: Re: [PATCH v1] crosvm: replace monolithic DEPENDS with PACKAGECONFIG flags
Date: Tue, 28 Jul 2026 10:37:30 -0700 (PDT)	[thread overview]
Message-ID: <6a68e8da.c40671ba.2ba795.19ab@mx.google.com> (raw)
In-Reply-To: <20260726202028.509096-1-kraghava@qti.qualcomm.com>

Thanks Keerthivasan — this cleanly addresses the TODO we left in v4 and
it's pretty close for a v1. A few questions before I take it; so let's
do them in a v2.

1. protobuf-native

   The old DEPENDS carried protobuf-native unconditionally, and the
   new base DEPENDS drops it without mapping it into any PACKAGECONFIG
   entry. Please confirm whether it's actually unused (in which case
   removing it as dead weight is fine — a one-liner in the commit
   message noting so would be great), or whether it's needed by a
   specific feature (in which case it should be attached to that
   feature's PACKAGECONFIG line, not left dangling).

   The test you ran ("default PACKAGECONFIG boots on x86-64") wouldn't
   necessarily exercise this — if some non-default feature (GPU,
   something else) pulls in protobuf codegen via a build.rs, that path
   is off in your test. A quick check with `--all-features` (or the
   full upstream default set) would answer it.

2. config-file in the default set

   Upstream crosvm's default features are approximately:

     ["audio", "balloon", "config-file", "document-features",
      "gpu", "libvda-stub", "net", "qcow", "slirp", "usb"]

   Your PACKAGECONFIG default is "audio balloon document-features qcow
   usb libvda-stub net slirp" — i.e. upstream's default minus gpu
   *and* minus config-file. GPU-off is deliberate and explained; the
   config-file omission isn't. Is that intentional (and if so, why)
   or an oversight? I'd lean toward matching upstream on this so our
   "default minus GPU" claim is exact.

3. document-features in the default set

   document-features is an attribute-macro crate whose sole purpose
   is generating documentation from feature declarations — it's a
   build-time doc helper, not a runtime feature. If this recipe isn't
   building or shipping docs by default (and it isn't), enabling it
   by default doesn't serve anyone. Please drop it from the default
   PACKAGECONFIG set. Keep it as a PACKAGECONFIG entry so someone
   who does want the doc build can flip it on.

I'll wait on a v2. Everything else looks good.

Bruce


      reply	other threads:[~2026-07-28 17:37 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-07-26 20:20 [PATCH v1] crosvm: replace monolithic DEPENDS with PACKAGECONFIG flags Keerthivasan Raghavan
2026-07-28 17:37 ` Bruce Ashfield [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=6a68e8da.c40671ba.2ba795.19ab@mx.google.com \
    --to=bruce.ashfield@gmail.com \
    --cc=meta-virtualization@lists.yoctoproject.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.