Closed tpetricek closed 8 years ago
Is the DropSparseRows behavior working in 1.2.4? Seems not to work... but I thought the fix went in before 1.2.4 went out?
NM, I see things work in 1.2.4. My projects are somehow still loading 1.2.3, configuration issue.
@adamklein Sorry for the mess here, I had this forgotten PR here. I suppose this is good to merge?
(This also now fixes the issue with CI builds.)
@adamklein Sorry, I made a bit of a mess here! This unmerged PR had an optimization that changed the code for reading CSV (to make it faster) - I adapted your changes to add Guid and DateTime support (but did it wrong the first time). Now it should all work again... we'll see if the CI finishes!
Okay, now we've got one genuine test failure in Can save MSFT data as CSV file and read it afterwards (with custom format)
on CI - this might be related to the DateTime support. I'll see what's going on.
The test failure was indeed caused by the DateTime
parsing.
Previously, we did not parse dates and so the test had an explicit call to DateTime.Parse
. Now, the date gets recognized & parsed automatically, so we do not need that. This commit changes the test to make it work.
@adamklein All seems to be green now, so let me know if you're OK with merging this!
Yes, let's merge! It's be great to release a minor version after these little fixes. Do you have any thoughts on the parenthesis issue in column names?
I changed how reading CSVs is done so that it does not allocate too much necessary memory. It is a bit faster and a bit less wasteful, but it is not as significant improvement as I'd be hoping for. More work would probably involve tweaking the CSV parser from F# Data.
I added test for
DropSparseRows
from @adamklein in #333. This seems to be working as expected. However, there was a recent(-ish) change in this bit (https://github.com/BlueMountainCapital/Deedle/commit/2a0fc636fb014432b56243870c5433a5fe59b50b), so perhaps you're not running the latest version?