feat: pin API #1045

* initial sweep through to understand how pin works. Did make some changes but mostly minor. * refactor pb schema to it's own file * fix: don't pin files during files.add if opts.pin === false * feat: add some http qs parsing, http route/resources cleanup, cleanup core/utils.parseIpfsPath * feat: expand pin tests. \nFirst draft. still needs some further work. * feat: Add logging for entry/exit of pins: add/rm/flush/load. Clean some documentation. * feat: add --pin to files.add, fix: improper pin option parsing in core. * feat: Use ipfs.files.add to add init-docs instead of directly using the unix-fs importer. * feat(tests): Add tests for cli --pin option. I know this should be more of an integration test. Should be written in /core. Maybe talk with Victor about testing different layers * feat: use isIPFS to valiate a multihash.

refactor: change pin.flush logging message

fix: don't need to cast the object.get result with toJSON revert: use interface-datastore.Key for datastore pin storage The proper change would be that datastore-level automatically casts operations into Keys fix: do not invoke callback within a try/catch feat(test): make cli pin tests more robust By using files that aren't added on IPFS initialization. Still needs work on files.rm (direct) and ipfs ls (indirect). fix: remove commented code, traced test failures to pin-set Got to go for the night, though, so will checkpoint here and address tomorrow. feat: parseIpfsPath now throws errors for consistency feat: resolveIpfsPaths error message lists the relative path that failed feat: use follow.bind instead of mutating the links Also decided not show relative paths. Less human friendly but probably cleaner otherwise. refactor: resolveIpfsPaths -> resolvePaths feat: promisify resolvePaths test: change parseIpfsPath failure tests to use try/catch docs: edit resolvePath doc revert: accidentally deleted commands/pin.js

I think my original rebase for this branch 2 weeks ago might have changed history for the intervening commits, indirectly causing some of these missed changes. or I just rebase onto the wrong oldparent. fix: some onlyHash and pin tests broke after merging onlyHash and pin interact: shouldn't pin when --only-hash. fix: trim output for 'pin ls when no hash is passed' test: indirect pins supersede direct pins: turns out we had a bug feat: add expectTimeout test utility feat: promisify some additional pin utils

I think I'll end up moving most tests here. test: add tests for pin.ls and pin.rm Based tests on other pin fixtures, need to migrate the isPinned* tests to them as well. fix: direct pins are now deleted by a default pin.rm(hash) test: prepare for pin.add tests 'indirect supersedes direct' test exposes a bug in pin.ls feat: switch away from multihashes for isPinned* tests test: impl pin.add tests fix: add fixture files only once test: add test for a potential bug, clean isPinned* tests refactor: remove a test that's no longer needed fix: pin.ls, indirect pins should supersede direct pins test: naive pin.load, pin.flush tests feat: remove most pin cli tests as functionality is tested in pin core tests refactor: rename solarSystem

fix: attempt to find a way to use http-api/inject test structure for pin tests test: fix pin.rm http-api tests test: fix pin.add http-api tests docs: docs and cleanup of http-api pin tests refactor: renaming fix: lint errors fix: resolvePaths tests are failing on CI, it might be long ops, testing a timeout bump fix: add files explicitly before testing resolvePaths fix: remove mocha.only from resolvePaths. let's hope tests pass, they are passing CI now fix: rename test/core/utils.spec.js -> utils.js so it's not run during browser tests

Need to leave computer, this is a checkpoint. test: add sanity test for walkItems and hasChild, clean others These tests are more descriptive than really pushing the impl. I'd love others' thoughts on what else should be hit and how. I also need to compare go's pinset impl against ours fix: stop daemons feat: documentation and multihash buffer handling for dag.get fix: lint

fix: pinset.hasChild buffer check feat: hardcode expected length for flush/load tests

I must have missed a commit during a rebase.

fix: yarg arugment naming for pin cli commands fix: convert file multihashes to a b58 string fix: another way of checking for CID-ness fix: lint fix: toB58String handles non-buffers fix: key-exchange core tests now shutdown daemon.

refactor: pinset.hasChild -> pinset.hasDescendent fix: invoke someCb if we've seen the hash before refactor: async patterns in dag._getRecursive refactor: pinset.hasDescendant refactor: pinset.storeItems async patterns refactor: pinset.loadSet and pin.walkItem async patterns docs: add link to go-ipfs' fanout bin implementation refactor: async patterns of pin.load/flush refactor: lint refactor: privatize internal pin key storage refactor: change encapsulation of ipfs.pin, fix resulting issues fix: lint fix: 'files add --pin=false' test was giving a false positive refactor: use is-ipfs to check CID-ability of a string refactor: remove last instance of 'once' in the pin code

They're simple enough, documented elsewhere, and not used by any exposed functionality.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: pin API #1045

feat: pin API #1045

Commits on Jun 18, 2018