Python has about 40 keywords, I say I would regularly use about 30, and irregularly use about another 5. Hardly seems like a "junkyard".
Further, this lack of first class support for lazy importing has spawned multiple CPython forks that implement their own lazy importing or a modified version of the prior rejected PEP 690. Reducing the real world need for forks seems worth the price of one keyword.
False await else import pass
None break except in raise
True class finally is return
and continue for lambda try
as def from nonlocal while
assert del global not with
async elif if or yield
Soft Keywords:
match case _ type
I think nonlocal/global are the only hard keywords I now barely use, for the soft ones I rarely use pattern matching, so 5 seems like a good estimate
> The choice to introduce a new `lazy` keyword reflects the need for explicit syntax. Lazy imports have different semantics from normal imports: errors and side effects occur at first use rather than at the import statement. This semantic difference makes it critical that laziness is visible at the import site itself, not hidden in global configuration or distant module-level declarations. The lazy keyword provides local reasoning about import behavior, avoiding the need to search elsewhere in the code to understand whether an import is deferred. The rest of the import semantics remain unchanged: the same import machinery, module finding, and loading mechanisms are used.
This functionality is highly desired, and it does appear to actually need a new (soft) keyword. Sorry you don't like it.
The pep didn’t mention considering reusing `async` instead of `lazy`. That would’ve conveyed the same thing to me without a new keyword, and would haven’t been similar to html’s usage `async`.
It is a 'soft keyword' as the PEP explains. I would not think that this has any major impact on anyone who just chooses to ignore this feature. Assuming that you want this behavior, I wonder how this could have been done in a better fashion without now having 'lazy' in the specific context of an import statement.
soft keyword for anyone not familiar like I was ...
"A new soft keyword lazy is added. A soft keyword is a context-sensitive keyword that only has special meaning in specific grammatical contexts; elsewhere it can be used as a regular identifier (e.g., as a variable name). The lazy keyword only has special meaning when it appears before import statements..."
> Python is quickly turning into a crowded keyword junkyard
* Javascript (ECMAScript) has 63 keywords.
* Rust has 50 keywords.
* Java has 51 keywords + 17 contextually reserved words, for a total of 68.
* Python has now 36 keywords + 4 'soft' keywords, for a total of 40.
* Go has 25 keywords.
Source phase imports are stage 3 (recommended for implementation, no major updates expected) in JS. "Import source" tells the runtime to go get the code, but not yet run it. Similar ideas seemingly to what's going on here in python!
https://github.com/tc39/proposal-source-phase-imports
This proposal defers even looking for the code. The LazyLoader already in the standard library would eagerly look for code, but just record a file path and not actually store any bytecode data, never mind deserializing or running it.
The rationale described in the PEP is that some systems try to `import`, for example, across a network share, so even searching the filesystem is slow and there is a desire to defer that (and avoid it on runs where the corresponding code isn't executed).
This one is scheduled to land in the next "minor" version, 3.15. Python has an annual release cadence; 3.14 came out recently and 3.15 is due next October.
In general, most PEPs are authored targeting the "next minor version" at the time of proposal; but they may be intentionally deferred at the start, and sometimes the process can take multiple years anyway.
There are also PEPs that don't involve any change to the Python language, standard library or interpreter. In particular, there are PEPs that exist simply to document existing practice (or changes thereto), PEPs that concern governance (the Python Software Foundation, the Steering Council etc.), and PEPs that cover related special interests, such as packaging standards (which in turn can range from technical details about how metadata is formatted, to changes in PyPI's API).
I'm confused, wouldn't HRT be happy about this? The article you linked specifically states "..we hope to propose a revised lazy imports PEP that introduces an explicit lazy keyword.."
Why would they? The accepted PEP just has extra keywords to mark imports explicit. That's a source level find-and-replace. They already did most of the real work in their codebase, such as finding where the code depended on the side effect of importing a module.
Dear PEP 810 authors. The Steering Council is happy to unanimously [4 votes, as Pablo cannot vote] accept “PEP 810, Explicit lazy imports”. Congratulations! We appreciate the way you were able to build on and improve the previously discussed (and rejected) attempt at lazy imports as proposed in PEP 690.
We have recommendations about some of the PEP’s details, a few suggestions for filling a couple of small gaps, and we have made decisions on the alternatives that you’ve left to the SC, all of which I’ll outline below. If you have any questions, please do reach out to the SC for clarification, either here, on the SC tracker, or in office hours.
Use lazy as the keyword. We debated many of the given alternatives (and some we came up with ourselves), and ultimately agreed with the PEP’s choice of the lazy keyword. The closest challenger was defer, but once we tried to use that in all the places where the term is visible, we ultimately didn’t think it was as good an overall fit. The same was true with all the other alternative keywords we could come up with, so… lazy it is!
What about from foo lazy import bar? Nope! We like that in both module imports and from-imports that the lazy keyword is the first thing on the line. It helps to visually recognize lazy imports of both varieties.
Leveraging a subclass of dict. We don’t see a need for this complicated alternative; please add this to the rejected ideas.
Allowing ’*’ in __lazy_modules__. We agree with the rationale for rejecting this idea; it can always be added later if needed.
One thing that the PEP does not mention is .pth files, which the site.py module processes, and which has some special handling for lines that begin with the string 'import' followed by a space or tab. It doesn’t make much sense for .pth files to support lazy imports, so we suggest that the PEP explicitly says that this special handling in .pth files will not be adapted to handle lazy imports.
There currently is no way to get the active filter mode, so please add a sys.get_lazy_imports() function. Also, do you think appending _mode to their names makes the purpose of these functions clearer? We leave that up to the PEP authors.
The PEP should be explicit about the precedence order between the different ways to set the mode, i.e. $PYTHON_LAZY_IMPORTS=<mode>, -X lazy_imports=<mode>, and sys.set_lazy_imports(). In all expectation, it will follow the same precedence order as other similar settings, but the PEP should be explicit.
We agree that the PEP should take no position on any style recommendations for sorting lazy imports. While we generally like the idea of grouping lazy imports together, let’s leave that up to the linters and auto-formatters to decide the details.
That should just about cover it. Again, thank you for your work on this, as it’s been a feature so many in the Python community have wanted for so long. Given the earlier attempts and existing workarounds, we think this strikes exactly the right balance.
Where <mode> can be:
"normal" (or unset): Only explicitly marked lazy imports are lazy
"all": All module-level imports (except in try blocks and import *) become potentially lazy
"none": No imports are lazy, even those explicitly marked with lazy keyword
When the global flag is set to "all", all imports at the global level of all modules are potentially lazy except for those inside a try block or any wild card (from ... import *) import.
This has been rejected several times before. Now that Google and Microsoft fired core developers, Python Inc. is owned by Meta. Meta supports lazy imports in Cinder and needs it for its atrocious and bloated scientific ecosystem so that the bloat loads in less than 5 min.
If Meta also starts firing Python developers, development might return to normal.
I've noticed that instead of defining requires at the top of Lua files, if you can know your own program well enough, defining them right before the dependency is actually used makes large Lua programs much, much quicker. Generally speaking, startup times can be a sped up by a meaningful factor.
I suspect this change in Python will dramatically improve the performance of such large programs as well.
I can't even express how negatively I feel about build/packaging systems that process dependencies based on code-level imports instead of some explicit build manifest separate to the code.
Or, you could just use a project-level specification file to list dependencies rather than looking for imports in the code, trying to figure out what they resolve to, and trying to package the results.
If you want even better nightmares, you can make localhost:8000 forward to a container running claude code with --dangerously-skip-permissions which uses an unkindness of mcp servers to control that endpoint on the fly based (amongst other sources) on 4chan's /b/.
I guess you meant "a feature of MCP servers which is unkind", but I couldn't help but interpret "unkindness" as the collective noun for a group of MCP servers.
Since this should be a rare thing I don't think it's unreasonable to require users of patterns like this to put some kind of special annotation for that static analysis tool saying "it may not look like it but I'm doing an import here".
Yeah, I'm aware the same behavior is available, but this proposal creates a call trigger on the dependency which requires far less analysis on larger projects to understand where the import needs to be moved to.
You have to create wrappers in languages like JavaScript, Lua, Python, etc. to create the same behavior.
Having some limited experience with lazy imports, yes, but this eliminates a lot of gross boilerplate. It also has the effect of "blessing" the practice of lazy imports which can have a cultural impact; it also prevents a situation wherein multiple subtly incompatible approaches to lazy imports become individually popular.
This is great for building modules. One can now lazy import all interesting names on __init__.py, so that instead of having to remember `from module.some_submodule_that_you_need_to_remember import method` you can just do `from module import name`.
> What about star imports (`from module import *`)?
> Wild card (star) imports cannot be lazy - they remain eager. This is because the set of names being imported cannot be determined without loading the module. Using the lazy keyword with star imports will be a syntax error. If lazy imports are globally enabled, star imports will still be eager.
Additionally, star imports can interfere with type checkers and IDEs and shadowing caused by star imports is a frequent and difficult to diagnose source of bugs (you analyze the function you think you're calling and find no issues, but you're actually calling a different function because the star import occurs after your explicit import).
You might be able to workaround this limitation by doing a lazy import into an intermediate module (a prelude) on a name by name basis and then star import that intermediate module. But personally I solve this problem using IDE features.
How will lazy imports interact with PEP 8 which recommends grouping imports in order:
1. Standard library imports.
2. Related third party imports.
3. Local application/library specific imports.
https://peps.python.org/pep-0008/#imports
Based on the examples in PEP 810, I suppose each group of regular imports can be followed by a group of lazy imports?
Python is quickly turning into a crowded keyword junkyard
Python has about 40 keywords, I say I would regularly use about 30, and irregularly use about another 5. Hardly seems like a "junkyard".
Further, this lack of first class support for lazy importing has spawned multiple CPython forks that implement their own lazy importing or a modified version of the prior rejected PEP 690. Reducing the real world need for forks seems worth the price of one keyword.
For those curious here are the actual keywords (from https://docs.python.org/3/reference/lexical_analysis.html?ut... )
Hard Keywords:
False await else import pass None break except in raise True class finally is return and continue for lambda try as def from nonlocal while assert del global not with async elif if or yield
Soft Keywords:
match case _ type
I think nonlocal/global are the only hard keywords I now barely use, for the soft ones I rarely use pattern matching, so 5 seems like a good estimate
I recall when they added "async" and it broken a whole lot of libraries. I hope they never again introduce new "hard" keywords.
Removing "print" in 3.0 helped their case significantly, as well.
From the PEP (https://peps.python.org/pep-0810/):
> The choice to introduce a new `lazy` keyword reflects the need for explicit syntax. Lazy imports have different semantics from normal imports: errors and side effects occur at first use rather than at the import statement. This semantic difference makes it critical that laziness is visible at the import site itself, not hidden in global configuration or distant module-level declarations. The lazy keyword provides local reasoning about import behavior, avoiding the need to search elsewhere in the code to understand whether an import is deferred. The rest of the import semantics remain unchanged: the same import machinery, module finding, and loading mechanisms are used.
This functionality is highly desired, and it does appear to actually need a new (soft) keyword. Sorry you don't like it.
The pep didn’t mention considering reusing `async` instead of `lazy`. That would’ve conveyed the same thing to me without a new keyword, and would haven’t been similar to html’s usage `async`.
I personally would have preferred "defer import os" instead of "lazy import os". It might be the non-native showing but lazy import feels unserious.
"Lazy" is standard language for this kind of behavior.
Lazy is more canonical: https://en.wikipedia.org/wiki/Lazy_evaluation
It is a 'soft keyword' as the PEP explains. I would not think that this has any major impact on anyone who just chooses to ignore this feature. Assuming that you want this behavior, I wonder how this could have been done in a better fashion without now having 'lazy' in the specific context of an import statement.
soft keyword for anyone not familiar like I was ...
"A new soft keyword lazy is added. A soft keyword is a context-sensitive keyword that only has special meaning in specific grammatical contexts; elsewhere it can be used as a regular identifier (e.g., as a variable name). The lazy keyword only has special meaning when it appears before import statements..."
> Python is quickly turning into a crowded keyword junkyard
* Javascript (ECMAScript) has 63 keywords. * Rust has 50 keywords. * Java has 51 keywords + 17 contextually reserved words, for a total of 68. * Python has now 36 keywords + 4 'soft' keywords, for a total of 40. * Go has 25 keywords.
Source link indicating PEP 810 was accepted:
https://discuss.python.org/t/pep-810-explicit-lazy-imports/1...
That was fast. It was sent to SC 20 days ago. (Not complaining. I am happy with the outcome).
Quite a few PEPs are accepted this quickly, actually.
But some others take literally years.
Source phase imports are stage 3 (recommended for implementation, no major updates expected) in JS. "Import source" tells the runtime to go get the code, but not yet run it. Similar ideas seemingly to what's going on here in python! https://github.com/tc39/proposal-source-phase-imports
This proposal defers even looking for the code. The LazyLoader already in the standard library would eagerly look for code, but just record a file path and not actually store any bytecode data, never mind deserializing or running it.
The rationale described in the PEP is that some systems try to `import`, for example, across a network share, so even searching the filesystem is slow and there is a desire to defer that (and avoid it on runs where the corresponding code isn't executed).
This will be huge at the place I work!
I’m unfamiliar with the PEP process. How long until this makes it into a Python version?
This one is scheduled to land in the next "minor" version, 3.15. Python has an annual release cadence; 3.14 came out recently and 3.15 is due next October.
In general, most PEPs are authored targeting the "next minor version" at the time of proposal; but they may be intentionally deferred at the start, and sometimes the process can take multiple years anyway.
There are also PEPs that don't involve any change to the Python language, standard library or interpreter. In particular, there are PEPs that exist simply to document existing practice (or changes thereto), PEPs that concern governance (the Python Software Foundation, the Steering Council etc.), and PEPs that cover related special interests, such as packaging standards (which in turn can range from technical details about how metadata is formatted, to changes in PyPI's API).
https://peps.python.org/pep-0000/
It should land in 3.15, so October next year. https://peps.python.org/pep-0790/
Given you can already trigger import in a local scope, it's merely a syntax sugar though.
Some folks at HRT[0] will probably be unhappy about that lol
0.https://www.hudsonrivertrading.com/hrtbeat/inside-hrts-pytho...
Look at the names in the PEP [1], this PEP is written by them
[1] https://peps.python.org/pep-0810/
I'm confused, wouldn't HRT be happy about this? The article you linked specifically states "..we hope to propose a revised lazy imports PEP that introduces an explicit lazy keyword.."
Is that not exactly what PEP 810 proposes?
Why would they? The accepted PEP just has extra keywords to mark imports explicit. That's a source level find-and-replace. They already did most of the real work in their codebase, such as finding where the code depended on the side effect of importing a module.
I think HN is translating the link somehow? It should be directing to this post: https://discuss.python.org/t/pep-810-explicit-lazy-imports/1...
"""
Dear PEP 810 authors. The Steering Council is happy to unanimously [4 votes, as Pablo cannot vote] accept “PEP 810, Explicit lazy imports”. Congratulations! We appreciate the way you were able to build on and improve the previously discussed (and rejected) attempt at lazy imports as proposed in PEP 690.
We have recommendations about some of the PEP’s details, a few suggestions for filling a couple of small gaps, and we have made decisions on the alternatives that you’ve left to the SC, all of which I’ll outline below. If you have any questions, please do reach out to the SC for clarification, either here, on the SC tracker, or in office hours.
Use lazy as the keyword. We debated many of the given alternatives (and some we came up with ourselves), and ultimately agreed with the PEP’s choice of the lazy keyword. The closest challenger was defer, but once we tried to use that in all the places where the term is visible, we ultimately didn’t think it was as good an overall fit. The same was true with all the other alternative keywords we could come up with, so… lazy it is!
What about from foo lazy import bar? Nope! We like that in both module imports and from-imports that the lazy keyword is the first thing on the line. It helps to visually recognize lazy imports of both varieties.
Leveraging a subclass of dict. We don’t see a need for this complicated alternative; please add this to the rejected ideas.
Allowing ’*’ in __lazy_modules__. We agree with the rationale for rejecting this idea; it can always be added later if needed.
One thing that the PEP does not mention is .pth files, which the site.py module processes, and which has some special handling for lines that begin with the string 'import' followed by a space or tab. It doesn’t make much sense for .pth files to support lazy imports, so we suggest that the PEP explicitly says that this special handling in .pth files will not be adapted to handle lazy imports.
There currently is no way to get the active filter mode, so please add a sys.get_lazy_imports() function. Also, do you think appending _mode to their names makes the purpose of these functions clearer? We leave that up to the PEP authors.
The PEP should be explicit about the precedence order between the different ways to set the mode, i.e. $PYTHON_LAZY_IMPORTS=<mode>, -X lazy_imports=<mode>, and sys.set_lazy_imports(). In all expectation, it will follow the same precedence order as other similar settings, but the PEP should be explicit.
We agree that the PEP should take no position on any style recommendations for sorting lazy imports. While we generally like the idea of grouping lazy imports together, let’s leave that up to the linters and auto-formatters to decide the details.
That should just about cover it. Again, thank you for your work on this, as it’s been a feature so many in the Python community have wanted for so long. Given the earlier attempts and existing workarounds, we think this strikes exactly the right balance.
-Barry, on behalf of the Python Steering Council
"""
HN is using the canonical URL for the page. See https://developer.mozilla.org/en-US/docs/Web/HTML/Attributes...
so it's a misconfig on python.org side?
I think so? It caused the same issue when I cross posted to Lobste.rs
[flagged]
Next we need
Is that not the purpose of the global switch outlined in the PEP?
No. From the PEP:
This has been rejected several times before. Now that Google and Microsoft fired core developers, Python Inc. is owned by Meta. Meta supports lazy imports in Cinder and needs it for its atrocious and bloated scientific ecosystem so that the bloat loads in less than 5 min.
If Meta also starts firing Python developers, development might return to normal.
I've noticed that instead of defining requires at the top of Lua files, if you can know your own program well enough, defining them right before the dependency is actually used makes large Lua programs much, much quicker. Generally speaking, startup times can be a sped up by a meaningful factor.
I suspect this change in Python will dramatically improve the performance of such large programs as well.
> I suspect this change in Python will dramatically improve the performance of such large programs as well.
Makes packaging super fun too, where you need to hit every possible path so you don't miss anything imported in 1% of the execution paths :)
I can't even express how negatively I feel about build/packaging systems that process dependencies based on code-level imports instead of some explicit build manifest separate to the code.
You and me both, but life as a consultant/freelancer requires you to drag yourself through dirt sometimes to make it out on the other side.
Not to mention the potential for runtime errors long after the code has started up.
Or, you could just use a project-level specification file to list dependencies rather than looking for imports in the code, trying to figure out what they resolve to, and trying to package the results.
Can't you do some kind of static analysis instead?
Depends if your code has horrors like this lurking in it:
m = importlib.import_module(requests.get("http://localhost:8000/package_name").content.strip().decode("ASCII"))
If you want even better nightmares, you can make localhost:8000 forward to a container running claude code with --dangerously-skip-permissions which uses an unkindness of mcp servers to control that endpoint on the fly based (amongst other sources) on 4chan's /b/.
> which uses an unkindness of mcp servers
I guess you meant "a feature of MCP servers which is unkind", but I couldn't help but interpret "unkindness" as the collective noun for a group of MCP servers.
That was indeed the intent. I first considered "a conspiracy" like with lemurs, but eventually felt that "unkindness" was more appropriate.
Better to let the viewers on a twitch stream vote for it.
Since this should be a rare thing I don't think it's unreasonable to require users of patterns like this to put some kind of special annotation for that static analysis tool saying "it may not look like it but I'm doing an import here".
Yes. It is much more clear to be explicit though.
Safer too.
You have always been able to do the same thing in Python. This PEP isn't needed for that functionality.
Yeah, I'm aware the same behavior is available, but this proposal creates a call trigger on the dependency which requires far less analysis on larger projects to understand where the import needs to be moved to.
You have to create wrappers in languages like JavaScript, Lua, Python, etc. to create the same behavior.
You can declare imports at the beginning of a program that don’t load until they are used?
"defining them right before the dependency is actually used"
Does it conform Occam's razor rule to have something that can be easily done very similar way without changing language?
Having some limited experience with lazy imports, yes, but this eliminates a lot of gross boilerplate. It also has the effect of "blessing" the practice of lazy imports which can have a cultural impact; it also prevents a situation wherein multiple subtly incompatible approaches to lazy imports become individually popular.
Not sure that's Occam's razor any more.
Regardless lazy loading needs widespread use to be most effective so having a unified syntax and no extra dependencies makes a lot of sense.
This is great for building modules. One can now lazy import all interesting names on __init__.py, so that instead of having to remember `from module.some_submodule_that_you_need_to_remember import method` you can just do `from module import name`.
https://peps.python.org/pep-0810/#what-about-star-imports-fr...
> What about star imports (`from module import *`)?
> Wild card (star) imports cannot be lazy - they remain eager. This is because the set of names being imported cannot be determined without loading the module. Using the lazy keyword with star imports will be a syntax error. If lazy imports are globally enabled, star imports will still be eager.
Additionally, star imports can interfere with type checkers and IDEs and shadowing caused by star imports is a frequent and difficult to diagnose source of bugs (you analyze the function you think you're calling and find no issues, but you're actually calling a different function because the star import occurs after your explicit import).
You might be able to workaround this limitation by doing a lazy import into an intermediate module (a prelude) on a name by name basis and then star import that intermediate module. But personally I solve this problem using IDE features.
https://github.com/python-lsp/python-lsp-server/blob/develop...
? This has nothing to do with star imports
My bad for misreading you