compose: Factor out helper for writing composejson #1636
Conversation
Prep for sharing this code with rojig.
|
|
||
| if (opt_write_composejson_to) | ||
| if (opt_write_commitid_to) |
There was a problem hiding this comment.
Hmm, I think this slightly changes the semantics of --write-commitid-to though. Before it would override ref, but now we're doing both. That's likely to cause issues in e.g. pungi, which does the ref update as a separate step.
There was a problem hiding this comment.
We have a test for that:
# --write-commitid-to should not set the ref
if ostree --repo=${repobuild} rev-parse ${treeref}; then
fatal "Found ${treeref} ?"
fi
echo "ok ref not written"
which is passing. I think the diff is just misleading. If you look at the logic above, we're still doing:
/* --write-commitid-to overrides writing the ref */
if (self->ref && !opt_write_commitid_to)
{
if (use_txn)
ostree_repo_transaction_set_ref (self->repo, NULL, self->ref, new_revision);
else
{
if (!ostree_repo_set_ref_immediate (self->repo, NULL, self->ref, new_revision,
cancellable, error))
return FALSE;
}
}
There was a problem hiding this comment.
Ahh yeah, you're right. I misread it. Although, this chunk now is no longer overridden by --write-commitid-to, right?
if (self->ref)
{
g_print ("%s => %s\n", self->ref, new_revision);
g_variant_builder_add (&composemeta_builder, "{sv}", "ref", g_variant_new_string (self->ref));
}So now, we're always printing that => line and inserting the ref key in the JSON even if we didn't actually update the ref, which is a bit misleading.
|
@rh-atomic-bot retry |
|
@rh-atomic-bot r=jlebon 84d2275 |
|
☀️ Test successful - status-atomicjenkins |
Prep for sharing this code with rojig.