block rm: change "force" option to "ignore"

[kevina]

"ignore" just seams like a more appropriate name as nothing is being
"forced"

License: MIT
Signed-off-by: Kevin Atkinson k@kevina.org

[kevina]

In the original pull request for block rm @jbenet suggested block rm -f as a way to be silent about deleting non-existent blocks (see #2962 (comment)). I take he was modeling that after the rm filestore utility. I want along with it, but never really liked that option name. This, p.r. changes the name to -i or --ignore. I think this is a better name because, unlike the unix rm command. the -f option does not force anything instead it "ignores" non-exist blocks rather than outputting an error.

@whyrusleeping thoughts?

[whyrusleeping]

I think that force is still the right wording for it. cc @jbenet @Kubuxu

[whyrusleeping]

'force' is the correct term here i think, we are forcing the removal of a block that we could otherwise not remove

[kevina]

[note, deleted last comment since it was inaccurate and it was better to start over]

@whyrusleeping

'force' is the correct term here i think, we are forcing the removal of a block that we could otherwise not remove

That is incorrect we are not forcing anything. All --force does is silence an error message when removing non-existent blocks. The end-result with and without the --force option is the same except for the error message and the exit code.

Reopening since i think there is confusion on what the --force option actually does. If we are clear on what it does and you still like --force better then we can agree to disagree and you can close this issue.

[whyrusleeping]

@kevina this is identical to unix rm in this case:

[whyrusleeping@rumil ~]$ rm foobar
rm: cannot remove 'foobar': No such file or directory
[whyrusleeping@rumil ~]$ echo $?
1
[whyrusleeping@rumil ~]$ rm -f foobar
[whyrusleeping@rumil ~]$ echo $?
0

[kevina]

Correct. Force also forces the removal of write protected files, this does not.

With that understanding if you still prefer --force go ahead and close this, just seams an odd name to me, that's all.

[whyrusleeping]

Thats true... Its not 'deleting a pinned block anyways'.

I'm just not sold on ignore... @Kubuxu @lgierth thoughts?

[ghost]

👎 Not a good enough reason to break existing interfaces. To us it looks like a simple rename, to a program the -f flag is just gone.

But I'm wondering why this has to fail on non-existent blocks in the first place. It could be properly idempotent. The result is the same (block non-existent), and if you need to know the block's state there's ipfs block stat. Unix's rm command can fail in various ways that you can --force your way around. block rm doesn't really have any failure modes that you can --force away.

[ghost]

And similarly, ipfs add doesn't fail if the stuff is already stored -- it's idempotent.

[ghost]

The only forceable thing I can imagine right now is rm'ing pinned blocks, but I don't think it matters much. We should never ever touch stuff that is pinned. Suddenly pinning would have to deal with blocks that might disappear, and that'd get very messy.

[kevina]

@lgierth the idempotent issue was discussed in #2962 and I agreed to use -f or -force for the idempotent behavior. I didn't think it was a very good name, but I didn't want to press the issue then. This p.r./issue exists because I hate to be stuck with a bad name.

[whyrusleeping]

I think even though ignore may be a marginally better name, we've already named it --force and breaking interfaces (even though nobody is probably using them yet) is generally frowned upon.

I'm gonna go ahead and close this