I'm sure with time DBOS will be great, I just did not have a lot of success with it when I tried it. It's quite complex, the quality of the SDKs was not overly amazing (when I initially used it, it had a ton of dependencies in it) and it just felt early.
Other question: why reimplementing your framework, rather than using an existing agent framework like Claude + MCP, or OpenAI + tool calling? Is it because you're using your own LM models, or just because you wanted more control on retries, etc?
There are not that many agent frameworks around at the moment. If you want to be provider independent you most likely either use pydantic AI or the vercel AI SDK would be my guess. Neither one have built-in solution for durable execution so you end up driving the loop yourself. So it's not that I don't use these SDKs, it's just that I need to drive the loop myself.
Okay very clear! I was saying that because your post example is just a kind of basic "tool use" example which is already implemented by MCP/OpenAI tool use, but obviously I guess your code can be suited to more complex scenarios
Two small questions:
1. in your README you give this example for durable execution:
I was just wondering, how does it work? I was more expecting a generator with a `yield` statement to run "long-running tasks" in the background... otherwise is the node runtime keeping the thread running with the await? doesn't this "pile up"?
2. would your framework be suited to long-running jobs with multiple steps? I have sometimes big jobs running in the background on all of my IoT devices, eg:
for each d in devices:
doSomeWork(d)
and I'd like to run the big outerloop each hour (say), but only if the previous one is complete (eg max num of workers per task = 1), and that the inner-loop be some "steps" that can be cached, but can be retried if they fail
would your framework be suited for that? or is that just a simpler use-case for pgmq and I don't need the Absurd framework?
> Okay very clear! I was saying that because your post example is just a kind of basic "tool use" example which is already implemented by MCP/OpenAI tool use, but obviously I guess your code can be suited to more complex scenarios
That's mostly just because I found that to be the easiest way to use any existing AI API to work. There are things like vercel's AI SDK which internally runs the agentic loop in generateText, but then there is no way to checkpoint that.
> I was just wondering, how does it work? I was more expecting a generator with a `yield` statement to run "long-running tasks" in the background... otherwise is the node runtime keeping the thread running with the await? doesn't this "pile up"?
When you `awaitEvent` or `sleepUntil`/`sleepFor` it sets a wake point or sets a re-schedule on the database. Then it raises `SuspendTask` and ends the execution of the task temporarily until it's rescheduled.
As for your IOT case: yes, you should be able to do that.
Oh interesting, maybe this makes for a better example then. If it has DBOS and Temporal it must be exposing some way to drive the loop. I'll investigate.
Really cool! How does it compare to DBOS ? https://docs.dbos.dev/architecture
I'm sure with time DBOS will be great, I just did not have a lot of success with it when I tried it. It's quite complex, the quality of the SDKs was not overly amazing (when I initially used it, it had a ton of dependencies in it) and it just felt early.
Other question: why reimplementing your framework, rather than using an existing agent framework like Claude + MCP, or OpenAI + tool calling? Is it because you're using your own LM models, or just because you wanted more control on retries, etc?
There are not that many agent frameworks around at the moment. If you want to be provider independent you most likely either use pydantic AI or the vercel AI SDK would be my guess. Neither one have built-in solution for durable execution so you end up driving the loop yourself. So it's not that I don't use these SDKs, it's just that I need to drive the loop myself.
Okay very clear! I was saying that because your post example is just a kind of basic "tool use" example which is already implemented by MCP/OpenAI tool use, but obviously I guess your code can be suited to more complex scenarios
Two small questions:
1. in your README you give this example for durable execution:
const shipment = await ctx.awaitEvent(`shipment.packed:${params.orderId}`);
I was just wondering, how does it work? I was more expecting a generator with a `yield` statement to run "long-running tasks" in the background... otherwise is the node runtime keeping the thread running with the await? doesn't this "pile up"?
2. would your framework be suited to long-running jobs with multiple steps? I have sometimes big jobs running in the background on all of my IoT devices, eg:
for each d in devices: doSomeWork(d)
and I'd like to run the big outerloop each hour (say), but only if the previous one is complete (eg max num of workers per task = 1), and that the inner-loop be some "steps" that can be cached, but can be retried if they fail
would your framework be suited for that? or is that just a simpler use-case for pgmq and I don't need the Absurd framework?
> Okay very clear! I was saying that because your post example is just a kind of basic "tool use" example which is already implemented by MCP/OpenAI tool use, but obviously I guess your code can be suited to more complex scenarios
That's mostly just because I found that to be the easiest way to use any existing AI API to work. There are things like vercel's AI SDK which internally runs the agentic loop in generateText, but then there is no way to checkpoint that.
> I was just wondering, how does it work? I was more expecting a generator with a `yield` statement to run "long-running tasks" in the background... otherwise is the node runtime keeping the thread running with the await? doesn't this "pile up"?
When you `awaitEvent` or `sleepUntil`/`sleepFor` it sets a wake point or sets a re-schedule on the database. Then it raises `SuspendTask` and ends the execution of the task temporarily until it's rescheduled.
As for your IOT case: yes, you should be able to do that.
Ah, got it, it throws Exception in order to stop the task each time https://github.com/earendil-works/absurd/blob/main/sdks/type...
> If you want to be provider independent you most likely either use pydantic AI ... Neither one have built-in solution for durable execution
PydanticAI has DBOS built in [0].
[0] https://ai.pydantic.dev/durable_execution/dbos/
Oh interesting, maybe this makes for a better example then. If it has DBOS and Temporal it must be exposing some way to drive the loop. I'll investigate.
Did Pydantic jump into both observability business (Logfire) and AI?
Yes! They have logfire and the AI framework (and they play nicely together and with DBOS).
https://docs.dbos.dev/integrations/logfire