Add an API to set/unset a deployment tree's mutability

[cgwalters]

This will be used by rpm-ostree to unset the immutable bit temporarily
in order to do package layering. We could add an API to deploy a tree
without the immutable bit, but this is simpler.

[mbarnes]

_ostree_linuxfs_alter_immutable_flag() doesn't really block and could drop its GCancellable parameter. Then you wouldn't need it for ostree_sysroot_deployment_set_mutable() either.
(I just find unnecessary GCancellable params to be misleading.)

Otherwise looks okay to me.

[cgwalters]

I guess I've found it easier so far to just carry them along everywhere, but I've no problem with dropping them either.

[mbarnes]

Just to note where I'm coming from on this -- on my old desktop project, blocking the user interface was a major problem. In reworking the APIs I used GCancellable as one way to clearly indicate which functions can block and may require an asynchronous version. I understand it's not as applicable here, but the tendency carries over.

[cgwalters]

To me this gets to a big conceptual question - e.g. opening a local file a blocking operation? What about directory enumeration? Gio at least includes GCancellable pretty much everywhere, and that's why I was too.

Perhaps a middle ground is that "single operations" like open() or unlink aren't cancellable, but enumerating a directory (or operating recursively on dirs, like glnx_shutil_rm_rf_at are cancellable even though they operate locally.

[mbarnes]

Right, multi-step operations have natural check points for cancellation, especially when the steps are repetitive. I guess my objection is more with taking a GCancellable argument but then not using it.

For something like _ostree_linuxfs_fd_alter_immutable_flag() it only really makes sense to check up front if the GCancellable is already cancelled before proceeding. But then I tend to think that should be left up to the caller.

[cgwalters]

I'm still waffling on this. If we added the stanza for checking the cancellable, i.e. match what e.g. g_file_delete does, would that address your objection?

The dual factors of

• Matching what Gio is doing
• Also given the fact that OSTree has so many APIs that all take cancellables, we'd create divergence by not doing so

have me leaning towards keeping them.

[mbarnes]

Didn't mean to make this into a big issue so I'll relax a bit.

I think as long as a function is just doing something with its GCancellable -- either checking it directly or passing it on -- I'm cool with keeping it. So yeah, the added stanza would address my objection.

[cgwalters]

I updated this to:

• Check the cancellable
• Use *at()