Permission to publish fork of inkwell

[Wodann]

We were in the process of publishing our crates, one of which depends on inkwell. We needed to make some changes to the original source (fork) and had hoped to be able to publish using the [patch.io-crates] functionality, but that is only valid for local builds.

It seems that the only alternative we really have is to publish our own version of inkwell as mun_inkwell, until such time as we can rely on a published version of inkwell. Would you approve of this? @TheDan64

[TheDan64]

The only change I see on your fork is the llvm version. Is it not possible for you to upgrade from 7 to 8?

[Wodann]

We were actually looking to update to LLVM 10, but the main problem is with what's not in our branch. After we forked, lifetimes were introduced to Inkwell. This requires a large refactor on our end, as some of our sub-system dependencies don't support lifetimes. We are working to roll out our own solution for those cases, but its non-trivial. As such we decided to not hold up the release on an LLVM update.

For now we've decided to release all of our crates that do not depend on inkwell and held off on ones that do. Going forward, the lack of progress on mutex features does pose a potential problem for our future releases. I'll move that discussion to #154

Thank you for all of your working on inkwell. It's been a huge help to our development! 🙂

[TheDan64]

To answer the issue more directly, I don't believe publishing a fork of inkwell is the appropriate solution.

We were actually looking to update to LLVM 10, but the main problem is with what's not in our branch. After we forked, lifetimes were introduced to Inkwell. This requires a large refactor on our end, as some of our sub-system dependencies don't support lifetimes. We are working to roll out our own solution for those cases, but its non-trivial. As such we decided to not hold up the release on an LLVM update.
For now we've decided to release all of our crates that do not depend on inkwell and held off on ones that do. Going forward, the lack of progress on mutex features does pose a potential problem for our future releases. I'll move that discussion to #154

Sorry to hear that you've had difficulty in updating. Unfortunately, adding lifetimes was necessary to ensure various objects didn't outlive the Context they spawned from (directly or even indirectly)

My personal recommendation would be to update your code to work with the lifetime changes, update to llvm8, and then publish with our llvm8 placeholder crate. Then, you should continue to develop pointing to our github branch for llvm10. We will likely publish another llvm placeholder crate (but for 10) at some point in the future.

Thank you for all of your working on inkwell. It's been a huge help to our development! 🙂

You're welcome! I'm thankful that we've had a lot of contributors who have helped us get this far.