Add a plugin to provide autoimport functionality

[bagel897]

Autoimport plugin for pylsp

Depends on accompanying pull request in rope
@bagel897
Closes#34

Prereqs

Rope > 1.0.0 (AKA git builds rn)

Demo Images

can detect external modules

grabs names from the external modules

Features

• Score-based sorting algorithm with thresholding. Picks the 25 best options.
  Will take suggestions on how to improve the basic algorithm
• Relatively high preformance using sqllite3 and ast.
  Besides the cache generation, it preforms fine on my machine
• Avoids suggestions on single line comments, from/normal imports, function/classes, dots
  Most of the test cases are to cover this and it should be pretty functional.
• Most completion related tests should be in rope
  the main ones are in parsing existing names and whether to suggest imports
• inserts all imports at the bottom of the set of imports. Recommendation is to use the isort extension.
• Watches for changes (but not deleted files) and regenerates cache for saved files.

Won't fix

1. Correct detection of type of statements from __init__.py modules. Would be difficult to preform type inference on.
   
2. Doesn't handle deleted files but this is a relatively rare.
3. Without a progress indicator, the user won't know when its ready (201) I have a WIP implementation based on Add progress reporting #236
4. Doesn't avoid suggestions on multiple line comments/strings, can fix if deemed important

[bagel897]

so this is almost ready:

1. It needs a fix for windows in rope Fix windows python-rope/rope#477 and associated release
2. Its still a little rough around where to provide suggestions and slow when it does. Ideally, some of this can be used in jedi and rope completions.
3. It generates a large cache which takes a while. Normally, this will be on-disk and I will work on improving it. But in CI, this takes a while and must be done seperately for each test and kept in memory. Either I change the scope of the associated fixtures to a module scope, or we have 2m long testing runs. In the future, I'll make it use configured dependencies automatically.

[bagel897]

Some notes about the future of this. I'm working on splitting it out to its own repository where 1-3 things will happen.

1. Rope-autoimport will use autoimport-core under the hood
2. Pylsp will use autoimport-core directly
3. Jedi will use autoimport-core, negating the requirements for any changes in pylsp.

But for now, this is perfectly functional code

[bagel897]

Yes, but its probably better to mark it off by default, and explain that it is experimental unless it is replaced in jedi/rope

[ccordoba12]

Agreed, that's fine for me too.

[mostley]

Can I help somehow to make this happen?

[bagel897]

Can I help somehow to make this happen?

If Jedi manifested a should_insert API, that'd help - but isn't necessary for the merge. I'm not sure what else is needed.

[ccordoba12]

I didn't know this was ready. I'll try to review it this week so we can include it in our next version.

[ccordoba12]

Thanks a lot @bagel897 for your work on this! I left a very lightweight review for you, mostly about style issues. Otherwise looks good to me.

[bagel897]

Pushed a bugfix and the style changes. Let me know if there's anything else (or if the CI fails, I'll check). You might want to squash the PR, since its a lot of poorly labeled commits, but up to you. Thanks for the feedback @ccordoba12!

[ccordoba12]

Thanks for addressing my previous review @bagel897! One last suggestion for you then this should be ready.

[ccordoba12]

You might want to squash the PR, since its a lot of poorly labeled commits, but up to you

Don't worry about that, we squash merge all PRs in this repo.

[ccordoba12]

Thanks @bagel897 for this addition! It's a great improvement!!

[bagel897]

Autoimport plugin for pylsp

Depends on accompanying pull request in rope
@bagel897
Closes#34

Prereqs

Rope > 1.0.0 (AKA git builds rn)

Demo Images

can detect external modules

grabs names from the external modules

Features

• Score-based sorting algorithm with thresholding. Picks the 25 best options.
  Will take suggestions on how to improve the basic algorithm
• Relatively high preformance using sqllite3 and ast.
  Besides the cache generation, it preforms fine on my machine
• Avoids suggestions on single line comments, from/normal imports, function/classes, dots
  Most of the test cases are to cover this and it should be pretty functional.
• Most completion related tests should be in rope
  the main ones are in parsing existing names and whether to suggest imports
• inserts all imports at the bottom of the set of imports. Recommendation is to use the isort extension.
• Watches for changes (but not deleted files) and regenerates cache for saved files.

Won't fix

1. Correct detection of type of statements from __init__.py modules. Would be difficult to preform type inference on.
   
2. Doesn't handle deleted files but this is a relatively rare.
3. Without a progress indicator, the user won't know when its ready (201) I have a WIP implementation based on Add progress reporting #236
4. Doesn't avoid suggestions on multiple line comments/strings, can fix if deemed important

[bagel897]

so this is almost ready:

1. It needs a fix for windows in rope Fix windows python-rope/rope#477 and associated release
2. Its still a little rough around where to provide suggestions and slow when it does. Ideally, some of this can be used in jedi and rope completions.
3. It generates a large cache which takes a while. Normally, this will be on-disk and I will work on improving it. But in CI, this takes a while and must be done seperately for each test and kept in memory. Either I change the scope of the associated fixtures to a module scope, or we have 2m long testing runs. In the future, I'll make it use configured dependencies automatically.

[bagel897]

Some notes about the future of this. I'm working on splitting it out to its own repository where 1-3 things will happen.

1. Rope-autoimport will use autoimport-core under the hood
2. Pylsp will use autoimport-core directly
3. Jedi will use autoimport-core, negating the requirements for any changes in pylsp.

But for now, this is perfectly functional code

[bagel897]

Yes, but its probably better to mark it off by default, and explain that it is experimental unless it is replaced in jedi/rope

[ccordoba12]

Agreed, that's fine for me too.

[mostley]

Can I help somehow to make this happen?

[bagel897]

Can I help somehow to make this happen?

If Jedi manifested a should_insert API, that'd help - but isn't necessary for the merge. I'm not sure what else is needed.

[ccordoba12]

I didn't know this was ready. I'll try to review it this week so we can include it in our next version.

[ccordoba12]

Thanks a lot @bagel897 for your work on this! I left a very lightweight review for you, mostly about style issues. Otherwise looks good to me.

[bagel897]

Pushed a bugfix and the style changes. Let me know if there's anything else (or if the CI fails, I'll check). You might want to squash the PR, since its a lot of poorly labeled commits, but up to you. Thanks for the feedback @ccordoba12!

[ccordoba12]

Thanks for addressing my previous review @bagel897! One last suggestion for you then this should be ready.

[ccordoba12]

You might want to squash the PR, since its a lot of poorly labeled commits, but up to you

Don't worry about that, we squash merge all PRs in this repo.

[ccordoba12]

Thanks @bagel897 for this addition! It's a great improvement!!

[bagel897]

Autoimport plugin for pylsp

Depends on accompanying pull request in rope
@bagel897
Closes#34

Prereqs

Rope > 1.0.0 (AKA git builds rn)

Demo Images

can detect external modules

grabs names from the external modules

Features

• Score-based sorting algorithm with thresholding. Picks the 25 best options.
  Will take suggestions on how to improve the basic algorithm
• Relatively high preformance using sqllite3 and ast.
  Besides the cache generation, it preforms fine on my machine
• Avoids suggestions on single line comments, from/normal imports, function/classes, dots
  Most of the test cases are to cover this and it should be pretty functional.
• Most completion related tests should be in rope
  the main ones are in parsing existing names and whether to suggest imports
• inserts all imports at the bottom of the set of imports. Recommendation is to use the isort extension.
• Watches for changes (but not deleted files) and regenerates cache for saved files.

Won't fix

1. Correct detection of type of statements from __init__.py modules. Would be difficult to preform type inference on.
   
2. Doesn't handle deleted files but this is a relatively rare.
3. Without a progress indicator, the user won't know when its ready (201) I have a WIP implementation based on Add progress reporting #236
4. Doesn't avoid suggestions on multiple line comments/strings, can fix if deemed important

[bagel897]

so this is almost ready:

1. It needs a fix for windows in rope Fix windows python-rope/rope#477 and associated release
2. Its still a little rough around where to provide suggestions and slow when it does. Ideally, some of this can be used in jedi and rope completions.
3. It generates a large cache which takes a while. Normally, this will be on-disk and I will work on improving it. But in CI, this takes a while and must be done seperately for each test and kept in memory. Either I change the scope of the associated fixtures to a module scope, or we have 2m long testing runs. In the future, I'll make it use configured dependencies automatically.

[bagel897]

Some notes about the future of this. I'm working on splitting it out to its own repository where 1-3 things will happen.

1. Rope-autoimport will use autoimport-core under the hood
2. Pylsp will use autoimport-core directly
3. Jedi will use autoimport-core, negating the requirements for any changes in pylsp.

But for now, this is perfectly functional code

[bagel897]

Yes, but its probably better to mark it off by default, and explain that it is experimental unless it is replaced in jedi/rope

[ccordoba12]

Agreed, that's fine for me too.

[mostley]

Can I help somehow to make this happen?

[bagel897]

Can I help somehow to make this happen?

If Jedi manifested a should_insert API, that'd help - but isn't necessary for the merge. I'm not sure what else is needed.

[ccordoba12]

I didn't know this was ready. I'll try to review it this week so we can include it in our next version.

[ccordoba12]

Thanks a lot @bagel897 for your work on this! I left a very lightweight review for you, mostly about style issues. Otherwise looks good to me.

[bagel897]

Pushed a bugfix and the style changes. Let me know if there's anything else (or if the CI fails, I'll check). You might want to squash the PR, since its a lot of poorly labeled commits, but up to you. Thanks for the feedback @ccordoba12!

[ccordoba12]

Thanks for addressing my previous review @bagel897! One last suggestion for you then this should be ready.

[ccordoba12]

You might want to squash the PR, since its a lot of poorly labeled commits, but up to you

Don't worry about that, we squash merge all PRs in this repo.

[ccordoba12]

Thanks @bagel897 for this addition! It's a great improvement!!

[bagel897]

Autoimport plugin for pylsp

Depends on accompanying pull request in rope
@bagel897
Closes#34

Prereqs

Rope > 1.0.0 (AKA git builds rn)

Demo Images

can detect external modules

grabs names from the external modules

Features

• Score-based sorting algorithm with thresholding. Picks the 25 best options.
  Will take suggestions on how to improve the basic algorithm
• Relatively high preformance using sqllite3 and ast.
  Besides the cache generation, it preforms fine on my machine
• Avoids suggestions on single line comments, from/normal imports, function/classes, dots
  Most of the test cases are to cover this and it should be pretty functional.
• Most completion related tests should be in rope
  the main ones are in parsing existing names and whether to suggest imports
• inserts all imports at the bottom of the set of imports. Recommendation is to use the isort extension.
• Watches for changes (but not deleted files) and regenerates cache for saved files.

Won't fix

1. Correct detection of type of statements from __init__.py modules. Would be difficult to preform type inference on.
   
2. Doesn't handle deleted files but this is a relatively rare.
3. Without a progress indicator, the user won't know when its ready (201) I have a WIP implementation based on Add progress reporting #236
4. Doesn't avoid suggestions on multiple line comments/strings, can fix if deemed important

[bagel897]

so this is almost ready:

1. It needs a fix for windows in rope Fix windows python-rope/rope#477 and associated release
2. Its still a little rough around where to provide suggestions and slow when it does. Ideally, some of this can be used in jedi and rope completions.
3. It generates a large cache which takes a while. Normally, this will be on-disk and I will work on improving it. But in CI, this takes a while and must be done seperately for each test and kept in memory. Either I change the scope of the associated fixtures to a module scope, or we have 2m long testing runs. In the future, I'll make it use configured dependencies automatically.

[bagel897]

Some notes about the future of this. I'm working on splitting it out to its own repository where 1-3 things will happen.

1. Rope-autoimport will use autoimport-core under the hood
2. Pylsp will use autoimport-core directly
3. Jedi will use autoimport-core, negating the requirements for any changes in pylsp.

But for now, this is perfectly functional code

[bagel897]

Yes, but its probably better to mark it off by default, and explain that it is experimental unless it is replaced in jedi/rope

[ccordoba12]

Agreed, that's fine for me too.

[mostley]

Can I help somehow to make this happen?

[bagel897]

Can I help somehow to make this happen?

If Jedi manifested a should_insert API, that'd help - but isn't necessary for the merge. I'm not sure what else is needed.

[ccordoba12]

I didn't know this was ready. I'll try to review it this week so we can include it in our next version.

[ccordoba12]

Thanks a lot @bagel897 for your work on this! I left a very lightweight review for you, mostly about style issues. Otherwise looks good to me.

[bagel897]

Pushed a bugfix and the style changes. Let me know if there's anything else (or if the CI fails, I'll check). You might want to squash the PR, since its a lot of poorly labeled commits, but up to you. Thanks for the feedback @ccordoba12!

[ccordoba12]

Thanks for addressing my previous review @bagel897! One last suggestion for you then this should be ready.

[ccordoba12]

You might want to squash the PR, since its a lot of poorly labeled commits, but up to you

Don't worry about that, we squash merge all PRs in this repo.

[ccordoba12]

Thanks @bagel897 for this addition! It's a great improvement!!

[bagel897]

Autoimport plugin for pylsp

Depends on accompanying pull request in rope
@bagel897
Closes#34

Prereqs

Rope > 1.0.0 (AKA git builds rn)

Demo Images

can detect external modules

grabs names from the external modules

Features

• Score-based sorting algorithm with thresholding. Picks the 25 best options.
  Will take suggestions on how to improve the basic algorithm
• Relatively high preformance using sqllite3 and ast.
  Besides the cache generation, it preforms fine on my machine
• Avoids suggestions on single line comments, from/normal imports, function/classes, dots
  Most of the test cases are to cover this and it should be pretty functional.
• Most completion related tests should be in rope
  the main ones are in parsing existing names and whether to suggest imports
• inserts all imports at the bottom of the set of imports. Recommendation is to use the isort extension.
• Watches for changes (but not deleted files) and regenerates cache for saved files.

Won't fix

1. Correct detection of type of statements from __init__.py modules. Would be difficult to preform type inference on.
   
2. Doesn't handle deleted files but this is a relatively rare.
3. Without a progress indicator, the user won't know when its ready (201) I have a WIP implementation based on Add progress reporting #236
4. Doesn't avoid suggestions on multiple line comments/strings, can fix if deemed important

[bagel897]

so this is almost ready:

1. It needs a fix for windows in rope Fix windows python-rope/rope#477 and associated release
2. Its still a little rough around where to provide suggestions and slow when it does. Ideally, some of this can be used in jedi and rope completions.
3. It generates a large cache which takes a while. Normally, this will be on-disk and I will work on improving it. But in CI, this takes a while and must be done seperately for each test and kept in memory. Either I change the scope of the associated fixtures to a module scope, or we have 2m long testing runs. In the future, I'll make it use configured dependencies automatically.

[bagel897]

Some notes about the future of this. I'm working on splitting it out to its own repository where 1-3 things will happen.

1. Rope-autoimport will use autoimport-core under the hood
2. Pylsp will use autoimport-core directly
3. Jedi will use autoimport-core, negating the requirements for any changes in pylsp.

But for now, this is perfectly functional code

[bagel897]

Yes, but its probably better to mark it off by default, and explain that it is experimental unless it is replaced in jedi/rope

[ccordoba12]

Agreed, that's fine for me too.

[mostley]

Can I help somehow to make this happen?

[bagel897]

Can I help somehow to make this happen?

If Jedi manifested a should_insert API, that'd help - but isn't necessary for the merge. I'm not sure what else is needed.

[ccordoba12]

I didn't know this was ready. I'll try to review it this week so we can include it in our next version.

[ccordoba12]

Thanks a lot @bagel897 for your work on this! I left a very lightweight review for you, mostly about style issues. Otherwise looks good to me.

[bagel897]

Pushed a bugfix and the style changes. Let me know if there's anything else (or if the CI fails, I'll check). You might want to squash the PR, since its a lot of poorly labeled commits, but up to you. Thanks for the feedback @ccordoba12!

[ccordoba12]

Thanks for addressing my previous review @bagel897! One last suggestion for you then this should be ready.

[ccordoba12]

You might want to squash the PR, since its a lot of poorly labeled commits, but up to you

Don't worry about that, we squash merge all PRs in this repo.

[ccordoba12]

Thanks @bagel897 for this addition! It's a great improvement!!

[bagel897]

Autoimport plugin for pylsp

Depends on accompanying pull request in rope
@bagel897
Closes#34

Prereqs

Rope > 1.0.0 (AKA git builds rn)

Demo Images

can detect external modules

grabs names from the external modules

Features

• Score-based sorting algorithm with thresholding. Picks the 25 best options.
  Will take suggestions on how to improve the basic algorithm
• Relatively high preformance using sqllite3 and ast.
  Besides the cache generation, it preforms fine on my machine
• Avoids suggestions on single line comments, from/normal imports, function/classes, dots
  Most of the test cases are to cover this and it should be pretty functional.
• Most completion related tests should be in rope
  the main ones are in parsing existing names and whether to suggest imports
• inserts all imports at the bottom of the set of imports. Recommendation is to use the isort extension.
• Watches for changes (but not deleted files) and regenerates cache for saved files.

Won't fix

1. Correct detection of type of statements from __init__.py modules. Would be difficult to preform type inference on.
   
2. Doesn't handle deleted files but this is a relatively rare.
3. Without a progress indicator, the user won't know when its ready (201) I have a WIP implementation based on Add progress reporting #236
4. Doesn't avoid suggestions on multiple line comments/strings, can fix if deemed important

[bagel897]

so this is almost ready:

1. It needs a fix for windows in rope Fix windows python-rope/rope#477 and associated release
2. Its still a little rough around where to provide suggestions and slow when it does. Ideally, some of this can be used in jedi and rope completions.
3. It generates a large cache which takes a while. Normally, this will be on-disk and I will work on improving it. But in CI, this takes a while and must be done seperately for each test and kept in memory. Either I change the scope of the associated fixtures to a module scope, or we have 2m long testing runs. In the future, I'll make it use configured dependencies automatically.

[bagel897]

Some notes about the future of this. I'm working on splitting it out to its own repository where 1-3 things will happen.

1. Rope-autoimport will use autoimport-core under the hood
2. Pylsp will use autoimport-core directly
3. Jedi will use autoimport-core, negating the requirements for any changes in pylsp.

But for now, this is perfectly functional code

[bagel897]

Yes, but its probably better to mark it off by default, and explain that it is experimental unless it is replaced in jedi/rope

[ccordoba12]

Agreed, that's fine for me too.

[mostley]

Can I help somehow to make this happen?

[bagel897]

Can I help somehow to make this happen?

If Jedi manifested a should_insert API, that'd help - but isn't necessary for the merge. I'm not sure what else is needed.

[ccordoba12]

I didn't know this was ready. I'll try to review it this week so we can include it in our next version.

[ccordoba12]

Thanks a lot @bagel897 for your work on this! I left a very lightweight review for you, mostly about style issues. Otherwise looks good to me.

[bagel897]

Pushed a bugfix and the style changes. Let me know if there's anything else (or if the CI fails, I'll check). You might want to squash the PR, since its a lot of poorly labeled commits, but up to you. Thanks for the feedback @ccordoba12!

[ccordoba12]

Thanks for addressing my previous review @bagel897! One last suggestion for you then this should be ready.

[ccordoba12]

You might want to squash the PR, since its a lot of poorly labeled commits, but up to you

Don't worry about that, we squash merge all PRs in this repo.

[ccordoba12]

Thanks @bagel897 for this addition! It's a great improvement!!

[bagel897]

Autoimport plugin for pylsp

Depends on accompanying pull request in rope
@bagel897
Closes#34

Prereqs

Rope > 1.0.0 (AKA git builds rn)

Demo Images

can detect external modules

grabs names from the external modules

Features

• Score-based sorting algorithm with thresholding. Picks the 25 best options.
  Will take suggestions on how to improve the basic algorithm
• Relatively high preformance using sqllite3 and ast.
  Besides the cache generation, it preforms fine on my machine
• Avoids suggestions on single line comments, from/normal imports, function/classes, dots
  Most of the test cases are to cover this and it should be pretty functional.
• Most completion related tests should be in rope
  the main ones are in parsing existing names and whether to suggest imports
• inserts all imports at the bottom of the set of imports. Recommendation is to use the isort extension.
• Watches for changes (but not deleted files) and regenerates cache for saved files.

Won't fix

1. Correct detection of type of statements from __init__.py modules. Would be difficult to preform type inference on.
   
2. Doesn't handle deleted files but this is a relatively rare.
3. Without a progress indicator, the user won't know when its ready (201) I have a WIP implementation based on Add progress reporting #236
4. Doesn't avoid suggestions on multiple line comments/strings, can fix if deemed important

[bagel897]

so this is almost ready:

1. It needs a fix for windows in rope Fix windows python-rope/rope#477 and associated release
2. Its still a little rough around where to provide suggestions and slow when it does. Ideally, some of this can be used in jedi and rope completions.
3. It generates a large cache which takes a while. Normally, this will be on-disk and I will work on improving it. But in CI, this takes a while and must be done seperately for each test and kept in memory. Either I change the scope of the associated fixtures to a module scope, or we have 2m long testing runs. In the future, I'll make it use configured dependencies automatically.

[bagel897]

Some notes about the future of this. I'm working on splitting it out to its own repository where 1-3 things will happen.

1. Rope-autoimport will use autoimport-core under the hood
2. Pylsp will use autoimport-core directly
3. Jedi will use autoimport-core, negating the requirements for any changes in pylsp.

But for now, this is perfectly functional code

[bagel897]

Yes, but its probably better to mark it off by default, and explain that it is experimental unless it is replaced in jedi/rope

[ccordoba12]

Agreed, that's fine for me too.

[mostley]

Can I help somehow to make this happen?

[bagel897]

Can I help somehow to make this happen?

If Jedi manifested a should_insert API, that'd help - but isn't necessary for the merge. I'm not sure what else is needed.

[ccordoba12]

I didn't know this was ready. I'll try to review it this week so we can include it in our next version.

[ccordoba12]

Thanks a lot @bagel897 for your work on this! I left a very lightweight review for you, mostly about style issues. Otherwise looks good to me.

[bagel897]

Pushed a bugfix and the style changes. Let me know if there's anything else (or if the CI fails, I'll check). You might want to squash the PR, since its a lot of poorly labeled commits, but up to you. Thanks for the feedback @ccordoba12!

[ccordoba12]

Thanks for addressing my previous review @bagel897! One last suggestion for you then this should be ready.

[ccordoba12]

You might want to squash the PR, since its a lot of poorly labeled commits, but up to you

Don't worry about that, we squash merge all PRs in this repo.

[ccordoba12]

Thanks @bagel897 for this addition! It's a great improvement!!

[bagel897]

Autoimport plugin for pylsp

Depends on accompanying pull request in rope
@bagel897
Closes#34

Prereqs

Rope > 1.0.0 (AKA git builds rn)

Demo Images

can detect external modules

grabs names from the external modules

Features

• Score-based sorting algorithm with thresholding. Picks the 25 best options.
  Will take suggestions on how to improve the basic algorithm
• Relatively high preformance using sqllite3 and ast.
  Besides the cache generation, it preforms fine on my machine
• Avoids suggestions on single line comments, from/normal imports, function/classes, dots
  Most of the test cases are to cover this and it should be pretty functional.
• Most completion related tests should be in rope
  the main ones are in parsing existing names and whether to suggest imports
• inserts all imports at the bottom of the set of imports. Recommendation is to use the isort extension.
• Watches for changes (but not deleted files) and regenerates cache for saved files.

Won't fix

1. Correct detection of type of statements from __init__.py modules. Would be difficult to preform type inference on.
   
2. Doesn't handle deleted files but this is a relatively rare.
3. Without a progress indicator, the user won't know when its ready (201) I have a WIP implementation based on Add progress reporting #236
4. Doesn't avoid suggestions on multiple line comments/strings, can fix if deemed important

[bagel897]

so this is almost ready:

1. It needs a fix for windows in rope Fix windows python-rope/rope#477 and associated release
2. Its still a little rough around where to provide suggestions and slow when it does. Ideally, some of this can be used in jedi and rope completions.
3. It generates a large cache which takes a while. Normally, this will be on-disk and I will work on improving it. But in CI, this takes a while and must be done seperately for each test and kept in memory. Either I change the scope of the associated fixtures to a module scope, or we have 2m long testing runs. In the future, I'll make it use configured dependencies automatically.

[bagel897]

Some notes about the future of this. I'm working on splitting it out to its own repository where 1-3 things will happen.

1. Rope-autoimport will use autoimport-core under the hood
2. Pylsp will use autoimport-core directly
3. Jedi will use autoimport-core, negating the requirements for any changes in pylsp.

But for now, this is perfectly functional code

[bagel897]

Yes, but its probably better to mark it off by default, and explain that it is experimental unless it is replaced in jedi/rope

[ccordoba12]

Agreed, that's fine for me too.

[mostley]

Can I help somehow to make this happen?

[bagel897]

Can I help somehow to make this happen?

If Jedi manifested a should_insert API, that'd help - but isn't necessary for the merge. I'm not sure what else is needed.

[ccordoba12]

I didn't know this was ready. I'll try to review it this week so we can include it in our next version.

[ccordoba12]

Thanks a lot @bagel897 for your work on this! I left a very lightweight review for you, mostly about style issues. Otherwise looks good to me.

[bagel897]

Pushed a bugfix and the style changes. Let me know if there's anything else (or if the CI fails, I'll check). You might want to squash the PR, since its a lot of poorly labeled commits, but up to you. Thanks for the feedback @ccordoba12!

[ccordoba12]

Thanks for addressing my previous review @bagel897! One last suggestion for you then this should be ready.

[ccordoba12]

You might want to squash the PR, since its a lot of poorly labeled commits, but up to you

Don't worry about that, we squash merge all PRs in this repo.

[ccordoba12]

Thanks @bagel897 for this addition! It's a great improvement!!