Centralize setting messageId in RequestCallback #57848
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This PR looks large, but all it does is add a dozen lines and remove a lot of other ones. One first advantage of using Futures is that we can easily chain some "post-processing" to them. Until now we needed to pass the message ID around everywhere because it was set separately by each method. Instead, we could simply add a follow-up step to the final future which sets this ID, and remove all the former logic. Differential Revision: [D28224477](https://our.internmc.facebook.com/intern/diff/D28224477/) [ghstack-poisoned]
lw
requested review from
H-Huang,
mingzhe09088,
mrshenli,
pritamdamania87,
rohan-varma,
wayi1 and
zhaojuanmao
as code owners
facebook-github-bot
added
oncall: distributed
💊 CI failures summary and remediationsAs of commit d564aa9 (more details on the Dr. CI page):
🕵️ 1 new failure recognized by patternsThe following CI failures do not appear to be due to upstream breakages:
|
| Job | Step | Action |
|---|---|---|
 pytorch_xla_linux_bionic_py3_6_clang9_build |
Build | 🔁 rerun |
This comment was automatically generated by Dr. CI (expand for details).
Follow this link to opt-out of these comments for your Pull Requests.Please report bugs/suggestions to the (internal) Dr. CI Users group.
This was referenced May 7, 2021
Closed
This PR looks large, but all it does is add a dozen lines and remove a lot of other ones. One first advantage of using Futures is that we can easily chain some "post-processing" to them. Until now we needed to pass the message ID around everywhere because it was set separately by each method. Instead, we could simply add a follow-up step to the final future which sets this ID, and remove all the former logic. Differential Revision: [D28224477](https://our.internmc.facebook.com/intern/diff/D28224477/) [ghstack-poisoned]
This PR looks large, but all it does is add a dozen lines and remove a lot of other ones. One first advantage of using Futures is that we can easily chain some "post-processing" to them. Until now we needed to pass the message ID around everywhere because it was set separately by each method. Instead, we could simply add a follow-up step to the final future which sets this ID, and remove all the former logic. Differential Revision: [D28224477](https://our.internmc.facebook.com/intern/diff/D28224477/) [ghstack-poisoned]
This was referenced May 17, 2021
mrshenli
reviewed
May 18, 2021
| return future; | ||
| } catch (std::exception& e) { | ||
| return asFuture(handleError(e, messageType, messageId)); | ||
| // Pass a dummy message ID since it will be overwritten anyways. |
There was a problem hiding this comment.
should we remove message id from handleError argument list?
There was a problem hiding this comment.
Please feel free to do this in followup PRs.
There was a problem hiding this comment.
My long-term idea is to remove the ID entirely from Message. I believe the ID is only used by the RPC agent to match requests with responses, so in a sense it's an "external" information (it shouldn't be inside the Message, it should be a "tag" that's attached to it).
| message->setId(id); | ||
| return message; | ||
| }, | ||
| c10::getCustomClassType<c10::intrusive_ptr<Message>>()); |
There was a problem hiding this comment.
IIUC, this does some type conversion + map lookup + branching. I wonder if it make sense to keel this c10::ClassTypePtr in Message implementation and reuse it in other places?
There was a problem hiding this comment.
I wouldn't worry about performance here. It's true that we need to do a map lookup and some bookkeeping, but only the first time, as the result is then cached for subsequent uses. See the static local variable here
pytorch/aten/src/ATen/core/ivalue.h
Lines 1161 to 1169 in df44f01
Moreover, since that function is templated, the type will be resolved at compile time. Hence a reasonable compiler will inline that code and this callsite will basically just become one memory access (plus a check if the cache has been initialized).
This PR looks large, but all it does is add a dozen lines and remove a lot of other ones. One first advantage of using Futures is that we can easily chain some "post-processing" to them. Until now we needed to pass the message ID around everywhere because it was set separately by each method. Instead, we could simply add a follow-up step to the final future which sets this ID, and remove all the former logic. Differential Revision: [D28224477](https://our.internmc.facebook.com/intern/diff/D28224477/) [ghstack-poisoned]
This PR looks large, but all it does is add a dozen lines and remove a lot of other ones. One first advantage of using Futures is that we can easily chain some "post-processing" to them. Until now we needed to pass the message ID around everywhere because it was set separately by each method. Instead, we could simply add a follow-up step to the final future which sets this ID, and remove all the former logic. Differential Revision: [D28224477](https://our.internmc.facebook.com/intern/diff/D28224477/) [ghstack-poisoned]
This PR looks large, but all it does is add a dozen lines and remove a lot of other ones. One first advantage of using Futures is that we can easily chain some "post-processing" to them. Until now we needed to pass the message ID around everywhere because it was set separately by each method. Instead, we could simply add a follow-up step to the final future which sets this ID, and remove all the former logic. Differential Revision: [D28224477](https://our.internmc.facebook.com/intern/diff/D28224477/) [ghstack-poisoned]
This was referenced May 20, 2021
This PR looks large, but all it does is add a dozen lines and remove a lot of other ones. One first advantage of using Futures is that we can easily chain some "post-processing" to them. Until now we needed to pass the message ID around everywhere because it was set separately by each method. Instead, we could simply add a follow-up step to the final future which sets this ID, and remove all the former logic. Differential Revision: [D28224477](https://our.internmc.facebook.com/intern/diff/D28224477/) [ghstack-poisoned]
dgl-intel
pushed a commit
to dgl-intel/pytorch
that referenced
this pull request
May 21, 2021
Pull Request resolved: pytorch#57848 This PR looks large, but all it does is add a dozen lines and remove a lot of other ones. One first advantage of using Futures is that we can easily chain some "post-processing" to them. Until now we needed to pass the message ID around everywhere because it was set separately by each method. Instead, we could simply add a follow-up step to the final future which sets this ID, and remove all the former logic. ghstack-source-id: 129567065 Differential Revision: [D28224477](https://our.internmc.facebook.com/intern/diff/D28224477/)
|
This pull request has been merged in 4ac18f6. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Stack from ghstack:
This PR looks large, but all it does is add a dozen lines and remove a lot of other ones.
One first advantage of using Futures is that we can easily chain some "post-processing" to them. Until now we needed to pass the message ID around everywhere because it was set separately by each method. Instead, we could simply add a follow-up step to the final future which sets this ID, and remove all the former logic.
Differential Revision: D28224477