-
Notifications
You must be signed in to change notification settings - Fork 71
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Replace objects with variants #553
Conversation
Did Jane Street cite any specific reasons? (Just out of curiosity.) |
I'm assuming Eio users should use |
Yes, that's right. It's a good idea anyway as you get extra checks and better compiler errors. |
1c1f35f
to
4d629ff
Compare
4c526e4
to
f6bde9a
Compare
Jane Street have requested that Eio not use objects. This commit switches to an alternative scheme for representing OS resources using variants instead. The changes for users of the library are minimal - only the types change. The exception to this is if you want to provide your own implementations of resources, in which case you now provide a module rather than a class. The (small) changes to the README give a good idea of the user-facing effect.
This is a continuation of ocaml-multicore#553.
This is a continuation of ocaml-multicore#553.
This is a continuation of ocaml-multicore#553.
This is a continuation of ocaml-multicore#553.
CHANGES: New features / API changes: - Replace objects with variants (@talex5 @patricoferris ocaml-multicore/eio#553 ocaml-multicore/eio#605 ocaml-multicore/eio#608, reviewed by @avsm). Some potential users found object types confusing, so we now use an alternative scheme for OS resources. For users of the resources, the only thing that changes is the types: - Instead of taking an argument of type `#foo`, you should now take `_ foo`. - Instead of returning a value of type `foo`, you should now return `foo_ty Eio.Resource.t`. To provide your own implementation of an interface, you now provide a module rather than an object. For example, to provide your own source flow, use `Eio.Flow.Pi.source (module My_source)`. If you want to define your own interfaces, see the `Eio.Resource` module documentation. - Add `Eio.Pool` (@talex5 @darrenldl ocaml-multicore/eio#602, reviewed by @patricoferris). A lock-free pool of resources. This is similar to `Lwt_pool`. - Add `Eio.Lazy` (@talex5 ocaml-multicore/eio#609, reviewed by @SGrondin). If one fiber tries to force a lazy value while another is already doing it, this will wait for the first one to finish rather than raising an exception (as `Stdlib.Lazy` does). - Add `Eio.Path.native` (@talex5 ocaml-multicore/eio#603, reviewed by @patricoferris). This is useful when interacting with non-Eio libraries, for spawning sub-processes, and for displaying paths to users. - Add `Flow.single_write` (@talex5 ocaml-multicore/eio#598). - Add `Eio.Flow.Pi.simple_copy` (@talex5 ocaml-multicore/eio#611). Provides an easy way to implement the `copy` operation when making your own sink. - Eio_unix: add FD passing (@talex5 ocaml-multicore/eio#522). Allows opening a file and passing the handle over a Unix-domain socket. - Add `Process.run ?is_success` to control definition of success (@SGrondin ocaml-multicore/eio#586, reviewed by @talex5). - Add `Eio_mock.Domain_manager` (@talex5 ocaml-multicore/eio#610). This mock domain manager runs everything in a single domain, allowing tests to remain deterministic. - Add `Eio.Debug.with_trace_prefix` (@talex5 ocaml-multicore/eio#610). Allows prefixing all `traceln` output. The mock domain manager uses this to indicate which fake domain is running. Bug fixes: - Fork actions must not allocate (@talex5 ocaml-multicore/eio#593). When using multiple domains, child processes could get stuck if they forked while another domain held the malloc lock. - eio_posix: ignore some errors writing to the wake-up pipe (@talex5 ocaml-multicore/eio#600). If the pipe is full or closed, the wake-up should simply be ignored. Build/test fixes: - Fix some MDX problems on Windows (@polytypic ocaml-multicore/eio#597). - The README depends on kcas (@talex5 ocaml-multicore/eio#606). - Clarify configuration for lib_eio_linux and enable tests on other arches (@dra27 ocaml-multicore/eio#592). - eio_linux tests: skip fixed buffer test if not available (@talex5 ocaml-multicore/eio#604). - eio_windows: update available line to win32 (@talex5 ocaml-multicore/eio#588 ocaml-multicore/eio#591).
CHANGES: New features / API changes: - Replace objects with variants (@talex5 @patricoferris ocaml-multicore/eio#553 ocaml-multicore/eio#605 ocaml-multicore/eio#608, reviewed by @avsm). Some potential users found object types confusing, so we now use an alternative scheme for OS resources. For users of the resources, the only thing that changes is the types: - Instead of taking an argument of type `#foo`, you should now take `_ foo`. - Instead of returning a value of type `foo`, you should now return `foo_ty Eio.Resource.t`. To provide your own implementation of an interface, you now provide a module rather than an object. For example, to provide your own source flow, use `Eio.Flow.Pi.source (module My_source)`. If you want to define your own interfaces, see the `Eio.Resource` module documentation. - Add `Eio.Pool` (@talex5 @darrenldl ocaml-multicore/eio#602, reviewed by @patricoferris). A lock-free pool of resources. This is similar to `Lwt_pool`. - Add `Eio.Lazy` (@talex5 ocaml-multicore/eio#609, reviewed by @SGrondin). If one fiber tries to force a lazy value while another is already doing it, this will wait for the first one to finish rather than raising an exception (as `Stdlib.Lazy` does). - Add `Eio.Path.native` (@talex5 ocaml-multicore/eio#603, reviewed by @patricoferris). This is useful when interacting with non-Eio libraries, for spawning sub-processes, and for displaying paths to users. - Add `Flow.single_write` (@talex5 ocaml-multicore/eio#598). - Add `Eio.Flow.Pi.simple_copy` (@talex5 ocaml-multicore/eio#611). Provides an easy way to implement the `copy` operation when making your own sink. - Eio_unix: add FD passing (@talex5 ocaml-multicore/eio#522). Allows opening a file and passing the handle over a Unix-domain socket. - Add `Process.run ?is_success` to control definition of success (@SGrondin ocaml-multicore/eio#586, reviewed by @talex5). - Add `Eio_mock.Domain_manager` (@talex5 ocaml-multicore/eio#610). This mock domain manager runs everything in a single domain, allowing tests to remain deterministic. - Add `Eio.Debug.with_trace_prefix` (@talex5 ocaml-multicore/eio#610). Allows prefixing all `traceln` output. The mock domain manager uses this to indicate which fake domain is running. Bug fixes: - Fork actions must not allocate (@talex5 ocaml-multicore/eio#593). When using multiple domains, child processes could get stuck if they forked while another domain held the malloc lock. - eio_posix: ignore some errors writing to the wake-up pipe (@talex5 ocaml-multicore/eio#600). If the pipe is full or closed, the wake-up should simply be ignored. Build/test fixes: - Fix some MDX problems on Windows (@polytypic ocaml-multicore/eio#597). - The README depends on kcas (@talex5 ocaml-multicore/eio#606). - Clarify configuration for lib_eio_linux and enable tests on other arches (@dra27 ocaml-multicore/eio#592). - eio_linux tests: skip fixed buffer test if not available (@talex5 ocaml-multicore/eio#604). - eio_windows: update available line to win32 (@talex5 ocaml-multicore/eio#588 ocaml-multicore/eio#591).
@talex5 As someone who is already using output : Eio_unix.sink I now have to write one of these: output : Eio_unix.sink_ty Eio.Std.r
output : Eio_unix.sink_ty Eio_unix.sink
output : 'a Eio_unix.sink (* and introduce a phantom variable *) The OO approach elegantly hides the difference between |
Let me comment hat i fully agree with what @favonia said above. This is certainly an API quality regression. |
What is your answer @talex5 to this API quality regression? |
From the PR:
Seems far more complicated than using simple objects ... |
CHANGES: New features / API changes: - Replace objects with variants (@talex5 @patricoferris ocaml-multicore/eio#553 ocaml-multicore/eio#605 ocaml-multicore/eio#608, reviewed by @avsm). Some potential users found object types confusing, so we now use an alternative scheme for OS resources. For users of the resources, the only thing that changes is the types: - Instead of taking an argument of type `#foo`, you should now take `_ foo`. - Instead of returning a value of type `foo`, you should now return `foo_ty Eio.Resource.t`. To provide your own implementation of an interface, you now provide a module rather than an object. For example, to provide your own source flow, use `Eio.Flow.Pi.source (module My_source)`. If you want to define your own interfaces, see the `Eio.Resource` module documentation. - Add `Eio.Pool` (@talex5 @darrenldl ocaml-multicore/eio#602, reviewed by @patricoferris). A lock-free pool of resources. This is similar to `Lwt_pool`. - Add `Eio.Lazy` (@talex5 ocaml-multicore/eio#609, reviewed by @SGrondin). If one fiber tries to force a lazy value while another is already doing it, this will wait for the first one to finish rather than raising an exception (as `Stdlib.Lazy` does). - Add `Eio.Path.native` (@talex5 ocaml-multicore/eio#603, reviewed by @patricoferris). This is useful when interacting with non-Eio libraries, for spawning sub-processes, and for displaying paths to users. - Add `Flow.single_write` (@talex5 ocaml-multicore/eio#598). - Add `Eio.Flow.Pi.simple_copy` (@talex5 ocaml-multicore/eio#611). Provides an easy way to implement the `copy` operation when making your own sink. - Eio_unix: add FD passing (@talex5 ocaml-multicore/eio#522). Allows opening a file and passing the handle over a Unix-domain socket. - Add `Process.run ?is_success` to control definition of success (@SGrondin ocaml-multicore/eio#586, reviewed by @talex5). - Add `Eio_mock.Domain_manager` (@talex5 ocaml-multicore/eio#610). This mock domain manager runs everything in a single domain, allowing tests to remain deterministic. - Add `Eio.Debug.with_trace_prefix` (@talex5 ocaml-multicore/eio#610). Allows prefixing all `traceln` output. The mock domain manager uses this to indicate which fake domain is running. Bug fixes: - Fork actions must not allocate (@talex5 ocaml-multicore/eio#593). When using multiple domains, child processes could get stuck if they forked while another domain held the malloc lock. - eio_posix: ignore some errors writing to the wake-up pipe (@talex5 ocaml-multicore/eio#600). If the pipe is full or closed, the wake-up should simply be ignored. Build/test fixes: - Fix some MDX problems on Windows (@polytypic ocaml-multicore/eio#597). - The README depends on kcas (@talex5 ocaml-multicore/eio#606). - Clarify configuration for lib_eio_linux and enable tests on other arches (@dra27 ocaml-multicore/eio#592). - eio_linux tests: skip fixed buffer test if not available (@talex5 ocaml-multicore/eio#604). - eio_windows: update available line to win32 (@talex5 ocaml-multicore/eio#588 ocaml-multicore/eio#591).
Jane Street and other users have requested that Eio not use objects. This PR proposes a new scheme for representing OS resources using variants instead. The changes for users of the library are minimal - only the types change. The exception to this is if you want to provide your own implementations of resources, in which case you now provide a module rather than a class. The (small) changes to the README give a good idea of the user-facing effect.
An
'a Resource.t
is a resource implementing at least the interfaces in'a
. e.g.[`Close | `Shutdown] Resource.t
is a resource that can be closed and shut down. This should also give slightly shorter error messages, since we only list interfaces, not all the methods and their types.I added
type 'a r = 'a Resource.t
toEio.Std
to keep the new types fairly short, plus various aliases. e.g. we have:With that, the signature of
Flow.copy
changes like this:Internally, there are lots of changes. For now, I have only updated the flow and socket types. These are the most complicated types as they have lots of sub-types (source, sink, close, shutdown, ro-file, rw-file, unix, etc), and are used in lots of places (random numbers, buffered reading and writing, processes). I think changing over the remaining types will be straight-forward.
As with objects, calling an operation does not require any allocation, and initial benchmarks show no noticable change in performance (unsurprisingly, since method dispatch is very fast compared to a system call). More optimisations are possible if necessary.