Re: [PATCH v2 1/2] net/slirp.c: Refactor address parsing

[Samuel Thibault <samuel.thibault@gnu.org>]

Hello,

Doug Evans, le lun. 08 févr. 2021 10:59:01 -0800, a ecrit:
> Samuel, how do qemu patches involving libslirp changes usually work?

Well, we haven't had many yet actually :)

> Should I have held off submitting the qemu patch until the libslirp
> prerequisite has been added to qemu's tree,

No, as this example shows, there are iterations that can happen on the
qemu side before we have something we can commit, so better do both in
parallel.

I don't know what qemu people think about the slirp submodule: do qemu
prefers to only track stable branchs, or would it be fine to track the
master branch? I guess you'd prefer to have a slirp stable release you
can depend on when releasing qemu, so perhaps better wait for a slirp
release before bumping in qemu, just to be safe? Which doesn't mean
avoiding submitting patchqueues that depend on it before that, better go
in parallel.

> or maybe I should include the libslirp patch so that people can at least apply
> it (with a caveat saying the patch is already in libslirp.git) until it's added
> to the qemu tree?

Not sure what is best here. We at least need something so that people
are not confused by the patchqueue calling some function that doesn't
exist in the submodule yet.

Samuel