feat: add support for Vensim's ALLOCATE BY PRIORITY function by NiharikaHari · Pull Request #809 · climateinteractive/SDEverywhere

NiharikaHari · 2026-04-20T07:27:52Z

Fixes #792

Feature: Allocate by priority #792

Adds support for Vensim’s allocate by priority function in the C runtime.
Mirrors the structure and style of the existing allocate_available implementation for consistency.
Includes Vensim’s example model (with .dat file) under models/ for validation and comparison of outputs.
Note: JS support is not implemented yet.

…here into feat/allocate-by-priority

chrispcampbell · 2026-04-21T02:03:49Z

@NiharikaHari: Thank you for your contribution! I have some high priority tasks to attend to for the next couple days, but I will try my best to review your work by Thursday.

chrispcampbell

Thank you @NiharikaHari for taking the time to contribute your work on this new function!

Most of my comments below are nitpicky things about code style. I haven't read the logic too closely for the _ALLOCATE_BY_PRIORITY C runtime function, but as long as it produces equivalent results to Vensim (and I can see that it does since the tests are passing), then I will trust that the logic works.

The test_js job is failing; that job does more than just run JS-related tests, it also runs the JS-level integration tests that exercise both C/Wasm and JS code gen, and there are Emscripten warnings flagged here (see build log) that will need to be resolved.

Also, please add at least one new test in gen-equation-c-from-vensim.spec.ts to verify that code gen works as expected. You can follow the patterns used in the existing should work for ALLOCATE AVAILABLE function (1D LHS, ...) test. Add your new test just after the last ALLOCATE AVAILABLE test in that file.

chrispcampbell · 2026-04-24T00:36:28Z

+
+  static double out_return[ALLOCATE_BY_PRIORITY_BUFSIZE];
+
+  fprintf(stderr, request_quantities);


We generally don't leave printf's in production code. I know that Todd included some printf's in the _ALLOCATE_AVAILABLE implementation above, which was fine because they are gated by #ifdef PRINT_ALLOCATIONS_DEBUG_INFO. So it's OK if you want to leave these fprintf calls here, but just make sure to wrap them in #ifdef PRINT_ALLOCATIONS_DEBUG_INFO / #endif blocks.

chrispcampbell · 2026-04-24T00:37:43Z

+    out_return[idx[i]] = out[i];
+  }
+
+  fprintf(stderr, out_return);


Same comment as above about removing this or putting it in an #ifdef block.

chrispcampbell · 2026-04-24T00:37:57Z

+
+  fprintf(stderr, out_return);
+  return out_return;
+


Remove these two blank lines.

chrispcampbell · 2026-04-24T00:38:03Z

+
+    // Reduce remaining supply
+    supply -= dx;
+


Remove blank line.

chrispcampbell · 2026-04-24T00:39:33Z

+  }
+
+  // Create the outputs array
+  for (size_t i = 0; i < num_requesters; i++) out_return[i] = 0.0;


I'm not a fan of single-line loops or conditionals. Make this:

for (size_t i = 0; i < num_requesters; i++) { out_return[i] = 0.0; }

chrispcampbell · 2026-04-24T00:39:56Z

+
+  // Continue allocating until supply is exhausted
+  while (supply > _epsilon) {
+


Remove blank line.

chrispcampbell · 2026-04-24T00:40:28Z

+        break;
+      }
+    }
+    if (!any_active) break;


Convert single-line conditional to multiple lines with braces.

chrispcampbell · 2026-04-24T00:40:37Z

+    // smallest of (next completion, next activation, remaining supply)
+    double dx = dx_next_top;
+
+    if (!isnan(dx_next_start) && dx_next_start < dx) dx = dx_next_start;


Convert single-line conditional to multiple lines with braces.

chrispcampbell · 2026-04-24T00:40:47Z

+    double dx = dx_next_top;
+
+    if (!isnan(dx_next_start) && dx_next_start < dx) dx = dx_next_start;
+    if (supply < dx) dx = supply;


Convert single-line conditional to multiple lines with braces.

chrispcampbell · 2026-04-24T00:43:36Z

          }
        }
        continue
+      } else if ((callExpr.fnId === '_ALLOCATE_BY_PRIORITY') & (index < 2)) {


There is a typo here: & should be &&.

Once you fix that, you should be able to remove the extra parentheses as they aren't needed.

Actually, I think you could probably just remove the index < 2 part here. Ideally we would have validation code here for each argument (similar to what the TODO comments say in the code that handles ALLOCATE AVAILABLE args above). But if we're not going to add that validation code now, I think you can drop the index < 2 part, otherwise it gives the impression that only those args need to be considered when really all args should get special treatment eventually.

NiharikaHari · 2026-04-24T01:49:52Z

Thank you @chrispcampbell for taking the time to review this. I'll be able to work on these changes next week. The logic in _ALLOCATE_BY_PRIORITY C runtime function mirrors the algorithm used in PySD.

travisfranck · 2026-04-24T11:17:49Z

@NiharikaHari This is a useful addition to SDE. Thank you for contributing! Does you model run now?

NiharikaHari · 2026-04-24T12:19:39Z

@travisfranck Yes, it does! Outside of this I'm doing some janky changes around setConstants and setLookups to read from CIN files and CSV respectively in C runtime. Excited to integrate SDE into our pipeline :)

NiharikaHari added 7 commits March 10, 2026 16:58

feat: generate code for allocate by priority

c659952

test: add example model for allocate by priority

4a65f7a

feat: Implemented algorithm for allocate by priority in C

6a4869a

Merge branch 'main' of https://github.com/climateinteractive/SDEveryw…

8b5913c

…here into feat/allocate-by-priority

feat: updated comments for allocate by priority

faaad83

test: Added spec and dat file to test allocate by priority

196369e

cleanup of code

0bbc3b5

chrispcampbell self-requested a review April 24, 2026 00:29

chrispcampbell requested changes Apr 24, 2026

View reviewed changes


		static double out_return[ALLOCATE_BY_PRIORITY_BUFSIZE];

		fprintf(stderr, request_quantities);


		// Continue allocating until supply is exhausted
		while (supply > _epsilon) {

Conversation

NiharikaHari commented Apr 20, 2026 • edited by chrispcampbell Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

chrispcampbell commented Apr 21, 2026

Uh oh!

chrispcampbell left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

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!

NiharikaHari commented Apr 24, 2026

Uh oh!

travisfranck commented Apr 24, 2026

Uh oh!

NiharikaHari commented Apr 24, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

NiharikaHari commented Apr 20, 2026 •

edited by chrispcampbell

Loading

chrispcampbell left a comment •

edited

Loading