Skip to content

Conversation

@asmeurer
Copy link
Contributor

Since you're cherrypicking, a new branch. Just one commit so far. I don't know if there will be more.

Several places in the code assume history objects have the strings attribute (and that it's a list). We could add strings as an abstractproperty on the abc, but then every history class would have to define strings as a property, which would generally require using a getter and a setter. I'm not sure if we want that level of complication, so I haven't done it here.
@codecov-io
Copy link

codecov-io commented May 10, 2018

Codecov Report

Merging #618 into 2.0 will decrease coverage by <.01%.
The diff coverage is 50%.

Impacted file tree graph

@@ Coverage Diff @@ ## 2.0 #618 +/- ## ========================================== - Coverage 70.91% 70.91% -0.01%  ========================================== Files 131 131 Lines 12161 12163 +2 ========================================== + Hits 8624 8625 +1  - Misses 3537 3538 +1
Impacted Files Coverage Δ
prompt_toolkit/history.py 59.52% <50%> (-0.24%) ⬇️

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update f255927...1b3f547. Read the comment docs.

@jonathanslenders
Copy link
Member

Thanks! This makes sense, but I think there is more to fix here.
I guess if we merge #554 then this fix is not required anymore.

The Xonsh people have also been looking at lazy history loading. I have to think about how to support that. I think maybe in general this API can be improved. I'm actually also not such a fan anymore of the use of __getitem__ in this case.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

3 participants