IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
441: docs: unlink matrix and discord, issues for questions r=aphelionz a=koivunej
unlink the channel as we haven't had much presence there in the recent months. as an alternative, list issues.
Co-authored-by: Joonas Koivunen <joonas.koivunen@gmail.com>
440: ci: stop running dht interop tests for cross targets r=koivunej a=koivunej
The DHT interop tests were accidentially ran also with the `matrix.platform.cross` targets, this stops it.
Co-authored-by: Joonas Koivunen <joonas.koivunen@gmail.com>
437: ci: no longer require review approvals to merge r=koivunej a=koivunej
This PR will allow `bors r+` before there are any review approvals. It will be good for smaller PRs like #436 or things which happen right before releasing. Alternatives include:
- to request and await for approvals (current situation)
- to merge without bors -- this is something I'd like to avoid doing
Now bors uses the PRs configuration so I don't require any approvals on this, but it has to come after #436 so that the workflow is fixed so at least this is gated on that :) So @aphelionz, @ljedrz, or anyone interested in this, how do you see this? I think the policy should still be to review all non-boring maintenance PRs.
I did check CONTRIBUTING.md and found that there is nothing regarding this.
I'll merge this at the earliest of tomorrow morning on +02:00.
Co-authored-by: Joonas Koivunen <joonas.koivunen@gmail.com>
439: Use Key-value DB for DataStore and PinStore r=koivunej a=fetchadd
This PR is for using key-value db like sled as DataStore and PinSotre, so that it is more efficient for pinning and easy to realize MFS.
The FsDatastore is inefficiency now, when checking a indirect pinned cid, all file in the FsDatastore will be visited and read, what more important is that it is hard to distinguish the data for pinning and the data for DataStore. The key-value db like sled will bring convenience and efficiency in the next work with its high efficiency and easy key/value operation. Leveldb may be the first choice of key-value dbs, but now there are no suitable rust leveldb crates; Sled is fast and simple, and it is usable on servers and phones from any C-compatible language.
This PR is the reopen of [PR434](https://github.com/rs-ipfs/rust-ipfs/pull/434). For there is some verbose commits like "replaceing leveldb with sled", the master branch of fetchadd/rust-ipfs in the PR434 is deprecated. Now the same changes are applied to the staging branch which is synced with the master of rust-ipfs/rust-ipfs. So Thanks for the PR of fetchadd/rust-ipfs from @koivunej which uses mutex to work around init(&self) and unsafe, but as the master is deprecated, I can only copy the code to staging branch.
Accorrding to the review from @ koivunej in [PR434](https://github.com/rs-ipfs/rust-ipfs/pull/434), the `pinstore_interface_tests` is added to KvDataStore, and now all tests are passed.
Closes#434.
Co-authored-by: fetchadd <kaizengliu@qq.com>
435: Clippy changes for 1.49, doc fixes r=koivunej a=koivunej
Address latest warnings with 1.48. It feels like we should have gotten the `some_str == ""` in earlier versions as well, other than this these are just minor idiomatic changes. Also includes:
- a previously forgotten panic documentation on ipfs-unxifs walker
- workflow fix to enable merges again (see #436)
Co-authored-by: Joonas Koivunen <joonas.koivunen@gmail.com>
bors behaviour changed some time ago w.r.t. having tasks with a single
name but one passed and one skipped. perhaps this will work without one
as well.
429: feat: repo locking with fs2 r=koivunej a=niklaslong
supersedes #426 and resolves#243.
This implements repo locking with fs2. The only concern for this crate is that it is poorly maintained but the functionality we need seems stable enough for now.
Certain tests are currently failing, notably because multiple nodes are being created with the same repo (dag tests for instance). Not sure what the solution to this will be...
Update: running tests with `--test-threads=1` passes all unit tests. The integration tests fail (I think because multiple nodes are being spun up with the same repo).
Update 2: all green, making sure test nodes use a different temp dir for the repo fixed the tests.
Co-authored-by: Niklas Long <niklas@equilibrium.co>
430: doc: update README r=koivunej a=niklaslong
What follows is just a suggestion. This updates the README with a _running the tests_ section, fixes the TOC, moves the _contributing_ section to where it's more obvious and links to the http crate tutorial.
Co-authored-by: Niklas Long <niklas@equilibrium.co>
