Add a --force flag to pin rm to ignore nonexistent pins and avoid bailing out

[schomatis]

No description provided.

[overbool]

@schomatis @kevina

In the original implementation, ipfs pin rm <invalid-hash> <valid-hash> <invalid-hash> will print Error: not pinned and stop unpinning the <valid-hash>. When we add --force option, Did you mean that we don't need to print Error: not pinned and continue to unpin <valid-hash>? If so, maybe we should modify the implementation about Unpin() in core/corerepo/pinning.go in the meantime. Because Unpin() will return error when it meet the first <invalid-hash>. Am I right?

[kevina]

I think unpin should be modified to continue to remove all valid pins. It should by default print an error on the invalid hash but continue anyway and then at the end print an error that not all pins can be removed at the end. If force is given it should ignore non-existent blocks but continue anyway and not return an error. In other words, the only think --force should do is not print or return an error.

ipfs block rm is a good model for this.

[schomatis]

It should by default print an error on the invalid hash but continue anyway

Yes, exactly. I didn't express myself correctly, I meant that the -f flag should not error out on missing pins (it can print to stderr or to the log the the pins it didn't find).

[kevina]

@schomatis what I am saying is that pin rm should not bail out even if the --force flag is not given.

[Stebalien]

So, forcing without a --force flag is a noticeable breaking change. How about:

1. Collecting the errors as @overbool is doing in cmds/pin: use coreapi/pin #5843.
2. If --force is not specified, emitting a final "errors occurred" error like we do in ipfs repo gc.

That way:

1. We don't break anything.
2. We still try to remove what we can.

---

We also need to involve the JS people. It looks like there was already a discussion here: https://github.com/ipfs/interface-ipfs-core/issues/127.

[Stebalien]

Let's move the discussion to https://github.com/ipfs/interface-ipfs-core/issues/127#issuecomment-456561650.