TY-1686 Analytics #54
Conversation
|
There seem to be a mismatch between CI cargo clippy and local cargo clippy I will look into this. |
|
I ran into this rust-lang/rust-clippy#4612, which is fixed but not yet on stable. I'm more surprised that I haven't ran into this before tbh., I guess mainly because I only ran clippy before releases in previous projects, for which I always also ran As a side note I noticed that currently clippy nightly warns about some committed code, we might want to fix that to prevent annoyances when that warnings land on stable. |
So the rust server has problems? |
Currently this is not implemented for initial ranking as it is currently not accessible. (Defaults to NaN.) Some parts are currently implemented in a performance wise suboptimal way.
Tests now use the test_ prefix and are not in sub-modules.
Unfortunately, this happens sometimes. It can also be due to a github action issue. Try to re-run the workflow 🙂 |
|
Thanks, I will. I anyway have to solve the minor merge conflict which popped up 😉 |
rustonaut
force-pushed
the
TY-1686-analytics
branch
from
64e62d1
to
0f2ced9
Compare
We talked about having a ci to check nightly but we didn't do it yet. In the ci we use a fixed rust version exactly to avoid that a new warning will suddenly break evreything. |
This includes a test of the `compute_analytics` methods, and a port of the test from the prev. dart implementation (as far as applicable).
- skip if no history available for a document - if we have no relevant historic information at all we do return an error - use fold in dcg - in pick highest skip search if we won't be able to insert it anyway - use Iterator::fuse() instead of FusedIterator
The new function is called nan_safe_f32_cmp.
|
There is one small commit missing, coming in less then 3min 😉 . |
assert!(approx_eq!(f32, left, right, ulps=n)) has the problem that it doesn't print information about the values when it fails so assert_f32_eq adds a format string which on failure will be used as error message.
|
Ok that was all. I already had used that helper in some places when the assertions failed |
References
Summary
initial rank(different PR)