zhr/http request consolidation #18

Gohlub · 2025-06-25T20:26:14Z

HTTP request routing:

Problem: All HTTP request were routed by JSON body de-serialization, which prevented simple GET requests from going through (and other methods which might not involve parameters).

Solution: a two-phase http routing system for HTTP requests.

In phase one, we first parse parameter-less HTTP requests (those without an argument in the signature) and order the handlers according to how specific they are (specific here meaning whether they have a defined method, path, both, either, or none). This part was a bit tricky since I kept getting errors where I would hit the wrong handler, but after exhaustive testing it seems to work. One important note here is that parameter-less HTTP handlers don't get added to the HPMRequest enum anymore.

Phase 2 is pretty much how it was done before:
Deserializes request body into HPMRequest enum and matches according to the function arguments.

I also made sure to include some more expressive error handling that touches other parts of the code (not strictly related to HTTP).

This PR is related to another PR on the kit WIT parser that I will make shortly.

nick1udwig

a few substantive possible changes, and a few smaller ones

nick1udwig · 2025-06-26T21:57:39Z

hyperprocess_macro/src/lib.rs

+                    current_path != "/" &&
+                    !current_path.starts_with("/assets") &&
+                    !current_path.starts_with("/static")
+                }


can we remove this special-casing of ui-related paths? E.g. see #19 as of 62e318a which seems to work fine without the special case

nick1udwig · 2025-06-26T22:03:25Z

hyperprocess_macro/src/lib.rs

+                        let mut ctx_mut = ctx.borrow_mut();
+                        ctx_mut.current_path = current_path;
+                        ctx_mut.current_http_method = current_method;
+                    });


is capturing and restoring context like this required? See 3568d35 which seems to work fine without. If we need it, please add a test that shows how removing this breaks

similar for below

is capturing and restoring context like this required? See 3568d35 which seems to work fine without. If we need it, please add a test that shows how removing this breaks

similar for below

in my testing, I noticed that without context preservation, even if the handler was routed properly, get_path() would return None, which is not ideal

nick1udwig · 2025-06-26T22:06:47Z

hyperprocess_macro/src/lib.rs

                            // Call the websocket handler if it exists
                            #ws_method_call

                            // Save state if needed
                            unsafe {
                                hyperware_app_common::maybe_save_state(&mut *state);
                            }
+                            hyperware_process_lib::logging::debug!("WebSocket message processed successfully");


Recommend removing this log

nick1udwig · 2025-06-26T22:07:40Z

hyperprocess_macro/src/lib.rs

@@ -1006,28 +1545,50 @@ fn generate_message_handlers(
                        // Save state if needed
                        hyperware_app_common::maybe_save_state(&mut *state);
                    }
+                    hyperware_process_lib::logging::debug!("Local message processed successfully");


recommend removing

nick1udwig · 2025-06-26T22:07:42Z

hyperprocess_macro/src/lib.rs

            // Process the local request based on our handlers (now including both local and remote handlers)
            match serde_json::from_slice::<HPMRequest>(message.body()) {
                Ok(request) => {
+                    hyperware_process_lib::logging::debug!("Successfully deserialized local request");


recommend removing

nick1udwig · 2025-06-26T22:17:35Z

hyperprocess_macro/src/lib.rs

+                                let mut ctx_mut = ctx.borrow_mut();
+                                ctx_mut.current_path = Some(current_path.clone());
+                                ctx_mut.current_http_method = Some(http_method.clone());
+                                hyperware_process_lib::logging::debug!("Set current_path to: {:?}", ctx_mut.current_path);


don't need both this and L1392: remove one

nick1udwig · 2025-06-26T22:20:22Z

README.md

+}
+```
+
+#### **UI Path Conflicts (Critical!):**


discussion in this section seems wrong even if we don't remove the ui path hardcoding

nick1udwig · 2025-06-26T22:22:06Z

README.md

+
+### Smart Routing System 
+
+The framework now uses intelligent priority-based routing that automatically chooses the best handler based on the request:


framework now uses

->

framework uses

nick1udwig · 2025-06-26T22:22:46Z

README.md

+// Set current_http_method to: Some("GET")
+```
+
+### Enhanced Error Handling (New!)


remove (New!)

nick1udwig · 2025-06-26T22:22:51Z

README.md

+
+### Enhanced Error Handling (New!)
+
+The framework now provides much more detailed error messages for debugging:


remove now

nick1udwig

More questions: why are we restricting parameter-less #[http]? This PR doesn't have any problems with your test cases (and some code I generated wants to use parameter-less HTTP posts): #20

nick1udwig · 2025-06-27T12:22:25Z

If we don't need UI special-casing, PR #19 could be merged as-is. If we want parameter-less #[http], PR #20 needs a little work (i.e. remove commented-out code; update some comments in the code)

…al-casing

Gohlub · 2025-06-27T15:28:41Z

More questions: why are we restricting parameter-less #[http]? This PR doesn't have any problems with your test cases (and some code I generated wants to use parameter-less HTTP posts): #20

I am not sure what you mean? The macro supports them well (as per the testing app I showed you) and I updated the WIT parser to handle them as well.

add response headers

use new process_lib

…esetting remove more path method resetting

add sleep

add persist on diff

…ss-requests try including parameterless requests

…nsolidation-testing Conflicts: hyperprocess_macro/src/lib.rs

…n-testing http request consolidation testing

Conflicts: hyperprocess_macro/src/lib.rs

nick1udwig · 2025-06-30T18:45:43Z

TODOs:

a RESPONSE_REGISTRY-like method for saving state that must survive an await
passing "long-lived HTTP request" test add long-lived test Gohlub/http-testing-ground#1
have thought about other long-lived request cases carefully and put state as appropriate

Gohlub added 7 commits June 24, 2025 11:46

merge develop into my branch

71ed53f

WIP

13ec28c

hotfix

e7a40a4

WIP

7f76df7

'working' version

605aade

working version

e9cd1d1

cargo fmt

2b20e97

Gohlub requested a review from nick1udwig June 25, 2025 20:26

Gohlub and others added 6 commits June 26, 2025 14:17

finally a version that acts according to expected behavior

b291813

final fixes

9d033eb

test GET ui hardcode removal

d7d686a

more aggressively remove ui special casing

0a44245

try even harder

62e318a

try removing setting and re-setting APP_HELPERS

3568d35

nick1udwig requested changes Jun 26, 2025

View reviewed changes

nick1udwig mentioned this pull request Jun 26, 2025

Zhr/wit hotfix hyperware-ai/kit#344

Open

nick1udwig added 3 commits June 26, 2025 16:49

try including parameterless requests

fd25127

try fix

550e25d

try again to fix

e8ca914

nick1udwig requested changes Jun 27, 2025

View reviewed changes

nick1udwig added 2 commits June 26, 2025 18:59

add response headers

16369af

persist OnDiff (i.e. only when state changes)

c206c6e

nick1udwig and others added 2 commits June 27, 2025 08:12

fix it

cf30695

Streamlined http path and method context management, removed UI speci…

505171a

…al-casing

nick1udwig added 4 commits June 27, 2025 10:09

use new process_lib

1cdbf54

remove more path/method re-setting

4a45e2a

remove un-setting current_path

7a645c5

set http-related context to None after done handling message

d5497f3

nick1udwig and others added 20 commits June 27, 2025 10:43

fix HashMap reference

c7e9e2b

borrow as mut

0b9ec32

remove a comment

724182e

add sleep()

cdd0e82

await the send in sleep

8d314c7

fix action & target

2f0c914

sleep: fix return type

67a3512

try to fix deserialization

2f1af9e

do repeat yourself

1edd7dc

fix typo

7fc30e0

Merge pull request #21 from hyperware-ai/hf/add-response-headers

b98de9a

add response headers

Merge pull request #22 from hyperware-ai/hf/use-new-process-lib

aa5019a

use new process_lib

Merge pull request #23 from hyperware-ai/hf/remove-more-path-method-r…

cb3ed74

…esetting remove more path method resetting

Merge pull request #24 from hyperware-ai/hf/add-sleep

989c4d1

add sleep

Merge pull request #26 from hyperware-ai/hf/add-persist-on-diff

c992f96

add persist on diff

Merge pull request #20 from hyperware-ai/hf/try-including-parameterle…

7e24c34

…ss-requests try including parameterless requests

Merge branch 'zhr/http-request-consolidation' into hf/http-request-co…

40a6f8c

…nsolidation-testing Conflicts: hyperprocess_macro/src/lib.rs

Merge pull request #19 from hyperware-ai/hf/http-request-consolidatio…

4a00168

…n-testing http request consolidation testing

Merge branch 'develop' into zhr/http-request-consolidation

85056e6

Conflicts: hyperprocess_macro/src/lib.rs

remove incorrect state reset

9836e2a

nick1udwig mentioned this pull request Jul 4, 2025

add http_method #14

Closed


		### Smart Routing System

		The framework now uses intelligent priority-based routing that automatically chooses the best handler based on the request:


		### Enhanced Error Handling (New!)

		The framework now provides much more detailed error messages for debugging:

zhr/http request consolidation #18

Are you sure you want to change the base?

zhr/http request consolidation #18

Uh oh!

Conversation

Gohlub commented Jun 25, 2025

Uh oh!

nick1udwig left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

nick1udwig left a comment

Choose a reason for hiding this comment

Uh oh!

nick1udwig commented Jun 27, 2025

Uh oh!

Gohlub commented Jun 27, 2025

Uh oh!

nick1udwig commented Jun 30, 2025

Uh oh!

Uh oh!