feat: add block "id" tracking to the interpreter

- update Block schema to add id field
- store the id in trace
- update react UI to use these values, rather than computing its own. this cleans up quite a bit.
- some initial def-use tracking via a `defsite` context field
- replace Dataflow topology visualization with simpler Memory tracking table

Signed-off-by: Nick Mitchell <[email protected]>

Author: starpit

pdl-live-react/demos/demo1.pdl

@@ -1,7 +1,7 @@
 description: Simple LLM interaction
 text:
 - "write a hello world example\n"
-- model: ollama/granite-code:8b
+- model: ollama/llama3.1:8b
   parameters:
     stop_sequences: '!'
     temperature: 0

pdl-live-react/demos/run.sh

@@ -1,14 +1,21 @@
-#!/bin/sh
+#!/usr/bin/env bash
 
 SCRIPTDIR=$(cd $(dirname "$0") && pwd)
 UI="$SCRIPTDIR"/.. # top of react UI
 TOP="$UI"/.. # top of repo
 T="$UI"/src/demos # place to store traces
 
-pdl --trace "$T"/demo1.json "$UI"/demos/demo1.pdl
-pdl --trace "$T"/demo2.json "$TOP"/examples/tutorial/model_chaining.pdl
-pdl --trace "$T"/demo3.json "$TOP"/examples/fibonacci/fib.pdl
-pdl --trace "$T"/demo4.json "$TOP"/examples/chatbot/chatbot.pdl # WARNING: this one requires some human interaction. TODO script this.
-pdl --trace "$T"/demo5.json "$TOP"/examples/talk/6-code-json.pdl
+pdl --trace "$T"/demo1.json <(cat "$UI"/demos/demo1.pdl | sed -E 's#(model: )(.+)#\1ollama/llama3.1:8b#g')
+pdl --trace "$T"/demo2.json <(cat "$TOP"/examples/tutorial/model_chaining.pdl | sed -E 's#(model: )(.+)#\1ollama/llama3.1:8b#g')
+pdl --trace "$T"/demo3.json <(cat "$TOP"/examples/fibonacci/fib.pdl | sed -E 's#(model: )(.+)#\1ollama/llama3.1:8b#g')
+pdl --trace "$T"/demo4.json <(cat "$TOP"/examples/chatbot/chatbot.pdl  | sed -E 's#(model: )(.+)#\1ollama/llama3.1:8b#g') <<EOF
+what is the fastest animal?
+no
+in europe?
+yes
+EOF
+cat "$TOP"/examples/talk/6-code-json.pdl | sed -E 's#(model: )(.+)#\1ollama/llama3.1:8b#g' > "$TOP"/examples/talk/6-code-json.pdl.tmp \
+    && pdl --trace "$T"/demo5.json "$TOP"/examples/talk/6-code-json.pdl.tmp \
+    && rm "$TOP"/examples/talk/6-code-json.pdl.tmp
 pdl --trace "$T"/demo6.json "$UI"/demos/error.pdl || true
-pdl --trace "$T"/demo7.json "$TOP"/examples/talk/4-function.pdl
+pdl --trace "$T"/demo7.json <(cat "$TOP"/examples/talk/4-function.pdl | sed -E 's#(model: )(.+)#\1ollama/llama3.1:8b#g')

pdl-live-react/package-lock.json

@@ -22,6 +22,7 @@
   "dependencies": {
     "@patternfly/react-code-editor": "^6.1.0",
     "@patternfly/react-core": "^6.1.0",
+    "@patternfly/react-table": "^6.1.0",
     "@patternfly/react-topology": "^6.1.0",
     "@tauri-apps/api": "^2",
     "@tauri-apps/plugin-cli": "^2.2.0",

pdl-live-react/package.json

@@ -1 +1 @@
-{"kind": "text", "start_nanos": 1738006438448127000, "end_nanos": 1738006439695520000, "timezone": "EST", "description": "Simple LLM interaction", "defs": {}, "text": ["write a hello world example\n", {"kind": "model", "context": [{"role": "user", "content": "write a hello world example\n"}], "start_nanos": 1738006438449110000, "end_nanos": 1738006439695499000, "timezone": "EST", "defs": {}, "platform": "litellm", "model": "ollama/granite-code:8b", "parameters": {"temperature": 0.0, "stop_sequences": "!"}, "result": "```python\nprint(\"hello world\")\n```\n"}], "result": "write a hello world example\n```python\nprint(\"hello world\")\n```\n"}
+{"kind": "text", "id": "text", "start_nanos": 1738862333350354000, "end_nanos": 1738862340857376000, "timezone": "EST", "description": "Simple LLM interaction", "defs": {}, "text": ["write a hello world example\n", {"kind": "model", "id": "text.1.model", "context": [{"role": "user", "content": "write a hello world example\n", "defsite": "text.0"}], "start_nanos": 1738862333351329000, "end_nanos": 1738862340857340000, "timezone": "EST", "defs": {}, "platform": "litellm", "model": "ollama/llama3.1:8b", "parameters": {"temperature": 0.0, "stop_sequences": "!"}, "result": "**Hello World Example**\n=======================\n\nBelow is a simple \"Hello, World!\" program in Python:\n\n```python\n# hello_world.py\n\ndef main():\n    \"\"\"Prints 'Hello, World!' to the console.\"\"\"\n    print(\"Hello, World!\")\n\nif __name__ == \"__main__\":\n    main()\n```\n\n**How to Run:**\n\n1. Save this code in a file named `hello_world.py`.\n2. Open your terminal or command prompt.\n3. Navigate to the directory where you saved the file.\n4. Type `python hello_world.py` and press Enter.\n\nYou should see \"Hello, World!\" printed on your console."}], "result": "write a hello world example\n**Hello World Example**\n=======================\n\nBelow is a simple \"Hello, World!\" program in Python:\n\n```python\n# hello_world.py\n\ndef main():\n    \"\"\"Prints 'Hello, World!' to the console.\"\"\"\n    print(\"Hello, World!\")\n\nif __name__ == \"__main__\":\n    main()\n```\n\n**How to Run:**\n\n1. Save this code in a file named `hello_world.py`.\n2. Open your terminal or command prompt.\n3. Navigate to the directory where you saved the file.\n4. Type `python hello_world.py` and press Enter.\n\nYou should see \"Hello, World!\" printed on your console."}

pdl-live-react/src/Context.ts

@@ -1 +1 @@
-{"kind": "text", "start_nanos": 1738006440779179000, "end_nanos": 1738006442322911000, "timezone": "EST", "description": "Model chaining", "defs": {}, "text": ["Hello\n", {"kind": "model", "context": [{"role": "user", "content": "Hello\n"}], "start_nanos": 1738006440780081000, "end_nanos": 1738006441153102000, "timezone": "EST", "defs": {}, "platform": "litellm", "model": "ollama/granite-code:8b", "parameters": {"stop_sequences": "!"}, "result": "Hi there! How can I assist you today?\n"}, "\nDid you just say Hello?\n", {"kind": "model", "context": [{"role": "user", "content": "Hello\n"}, {"content": "Hi there! How can I assist you today?\n", "role": "assistant"}, {"role": "user", "content": "\nDid you just say Hello?\n"}], "start_nanos": 1738006441154069000, "end_nanos": 1738006442322886000, "timezone": "EST", "defs": {}, "platform": "litellm", "model": "ollama/granite-code:8b", "parameters": {"stop_sequences": "!"}, "result": "Yes, that's correct. As an AI language model, my goal is to provide helpful and accurate responses to your questions and conversations. I'm here to assist you in any way I can. So, what can I help you with today?\n"}], "result": "Hello\nHi there! How can I assist you today?\n\nDid you just say Hello?\nYes, that's correct. As an AI language model, my goal is to provide helpful and accurate responses to your questions and conversations. I'm here to assist you in any way I can. So, what can I help you with today?\n"}
+{"kind": "text", "id": "text", "start_nanos": 1738862342134373000, "end_nanos": 1738862343556101000, "timezone": "EST", "description": "Model chaining", "defs": {}, "text": ["Hello\n", {"kind": "model", "id": "text.1.model", "context": [{"role": "user", "content": "Hello\n", "defsite": "text.0"}], "start_nanos": 1738862342135343000, "end_nanos": 1738862342794512000, "timezone": "EST", "defs": {}, "platform": "litellm", "model": "ollama/llama3.1:8b", "parameters": {"stop_sequences": "!"}, "result": "Hello! Is there something I can help you with or would you like to chat?"}, "\nDid you just say Hello?\n", {"kind": "model", "id": "text.3.model", "context": [{"role": "user", "content": "Hello\n", "defsite": "text.0"}, {"content": "Hello! Is there something I can help you with or would you like to chat?", "role": "assistant", "defsite": "text.1.model"}, {"role": "user", "content": "\nDid you just say Hello?\n", "defsite": "text.2"}], "start_nanos": 1738862342795067000, "end_nanos": 1738862343556032000, "timezone": "EST", "defs": {}, "platform": "litellm", "model": "ollama/llama3.1:8b", "parameters": {"stop_sequences": "!"}, "result": "Yes, I did say hello. It's a greeting, typically used when starting a conversation or interaction. What's on your mind?"}], "result": "Hello\nHello! Is there something I can help you with or would you like to chat?\nDid you just say Hello?\nYes, I did say hello. It's a greeting, typically used when starting a conversation or interaction. What's on your mind?"}