feat(sessions): sessions table on project page #5204
Conversation
dosubot
bot
added
the
size:XXL
RogerHYang
changed the title
|
Looks good! Most of my comments are about optimizations for dataloaders and avoiding pulling from attributes over a connection. Assuming we will address these as follow-ups. Would be good to get a frontend review here as well. |
src/phoenix/db/migrations/versions/4ded9e43755f_create_project_sessions_table.py
Show resolved
Hide resolved
| models.Span.attributes[INPUT_VALUE].label("value"), | ||
| models.Span.attributes[INPUT_MIME_TYPE].label("mime_type"), |
There was a problem hiding this comment.
Non-blocking: We might want to think about materializing the input and output as their own columns so we can avoid querying attributes over a connection. Even with a dataloader, this led to performance issues with traces when the attributes are large.
|
|
||
| async def _load_fn(self, keys: List[Key]) -> List[Result]: | ||
| stmt = select(models.Trace).where(models.Trace.trace_id.in_(keys)) | ||
| async with self._db() as session: |
There was a problem hiding this comment.
ahh the db sessions plus the product sessions, they have arrived haha
| rootSpan { | ||
| input { | ||
| value | ||
| } | ||
| output { | ||
| value |
There was a problem hiding this comment.
yeah i see your point here, if there isn't a root span, this is gonna get nasty to deal with, can cross that bridge when we come to it i suppose
resolves #4959
resolves #5011
resolves #5014
resolves #5058
resolves #5059
resolves #5066