improve verbose of ChunkRecordingExecutor#2795
Conversation
h-mayorquin
commented
May 2, 2024

for more information, see https://pre-commit.ci
|
Is this helpful to the end-user or to the developer? This seems more like a verbose level 2 rather than a verbose level 1 type thing to me. But since this library only has two options (nothing + everything) I guess this could work. Who do you envision using this? |
|
@zm711 Users actually. I was thinking that people might want to know how much memory they will be using in total. That's personally what I want to know when I am running a task. When I developed I use a memory profiler not this. Maybe I am very off base about the need of the users, how are you thinking about it? |
|
I think I'm thinking about it from the basic biologist side. I might set memory limits in the function, but I don't really care if it is 950 vs 980 if I asked for 1GB. Even chunk_size doesn't really matter. If I request 1 sec and then it says 30000 (because that is my sampling_rate) I'm like okay cool. I think this could be useful for debugging some users when they run into multiprocessing issues, but for me my concern is how much info is too much info for the user (that's why for this thing I would prefer a verbosity level. So users that really need to know this info can opt-in to more and more information). But I'm not against this I just think there will be a subset of users who don't really care, but still want verbose=True to make sure the program is running. |
|
I don't understand, right now the My confusion is that I don't see how getting more information about the paralleling processing parameters (n_jobs vs n_jobs, memory, chunk_duration, etc) changes things qualitatively. As in,
I am not sure the current |
|
I think you're right. I don't actually see the value of verbose at all for this. I think tqdm is nice, but the other info is not needed so for me my hierarchy would be verbose=0: Nothing But since that isn't I withdraw my question :) |
|
Yeah, sometimes So from your perspective, here we should have |
|
Yeah, I think the information here is useful for people that need it but most users don't care about the inner working of the chunking so default don't give it to them and if someone wants it make them request it. |
|
That makes sense to me. It seems that is |
| chunk_duration_str = convert_seconds_to_str(chunk_duration) | ||
| print( | ||
| self.job_name, | ||
| "\n" |
There was a problem hiding this comment.
not sure 2 lines is a good idea because with the progressbar this will be 3 lines in total.
There was a problem hiding this comment.
I am fine with reducing it to two. Any preferences @alejoe91 ?
|
This is cool. I think I prefer the one line version. |
