దామోదర

Current content-cache layer as it's existing works well only in non-concurrent contexts. If another process updates object, cache will contain stale entry for ever. If we can have #825 ,...

Size hint can be useful to efficiently allocate memory in few cases, instead of exact size. For e.g. hyper uses [`SizeHint`](https://docs.rs/hyper/latest/hyper/body/struct.SizeHint.html) for specifying size hints in it's content body.

Can you specify, on how one have to proceed, if they have a stream of unknown length?

> > In the future, OpenDAL will support multipart uploads like #256 That would be great and may be very required for abstraction layer like OpenDAL. We on many cases...

🙏 And thanks for great work. OpenDAL will enable us to build a robust [solid](https://solidproject.org/) server, that can use any of object stores as backend.

Yes. This sounds much better. I am thinking of opening an issue for recoding last-modified. `Inode` just works for that.

Hello @Xuanwo , `rsfs` crate includes a very solid and well tested implementation of inmemory concurrent file system, that follows unix semantics(including inodes, rwx permissions, etc.). Here is code (MIT)....

SeeAlso: , if the inmem implementation can be maintained instead of redoing everything again.

@beebs-systap , @iansrobinson et al., Can support for virtual graphs which is there in blazegraph versions be available in neptune? It solves many practical problems. And seems easy to implement....

Yes. And much better is, the specific-resource id to be used in replacement for both `id` and `region` together if possible. Because, `SpecificResource` (in this case) by definition contains information...