Correctly handle ANSI Escape Codes in ElideMiddle, fix #713 #2428
Conversation
jhasse
force-pushed
the
elide-middle-ansi-escape
branch
from
5c960f9 to
5c253c4
Compare
|
Ha, looking at your diff I think I just had it wrong. EIther an off-by-one or maybe a rounding mistake in the |
|
elide_size rounds down and if the width is even (let's say 12), this means that the result would be:
To fix this we'd have to show another character either in the beginning or the end (my preference). I think this only works with an explicit If you think this was unintentional I would have a go at it in another PR :) |
jhasse
force-pushed
the
elide-middle-ansi-escape
branch
from
5c253c4 to
7454b0b
Compare
There was a problem hiding this comment.
Ok, playing around with it more, it seems to not work quite right. Here is a copy of the tmux buffer on a pane that is narrow enough to require elision (executed from with nvim, but the elision code is definitely involved here).
`tmux` buffer state
[-Wunused-variable] 3577 | FlagClassification cls = FlagClassification::PrivateFlag; [-Wunused-variable] 3582 | FlagClassification cls = FlagClassification::LocationFlag; [-Wunused-variable] 3583 | FlagKind kind = FlagKind::BuildSystem; 0-5->13/[email protected]] Linking CXX executable bin/ctest | ^~~~ 0-4->14/[email protected]] Linking CXX executable bin/cmake [15-1->2/[email protected]] Linking CXX static library Source/libCMakeLib.a 0-3->15/[email protected]] Linking CXX executable bin/ccmake [5-10->3/[email protected]] Linking CXX executable Tests/CMakeLib/testAffinity 0-2->16/[email protected]] Linking CXX executable bin/ctresalloc [4-10->4/[email protected]] Linking CXX executable Tests/CMakeLib/testUVProcessChainHelper 0-1->17/[email protected]] Linking CXX executable Tests/CMakeLib/CMakeLibTests [3-10->5/[email protected]] Linking CXX executable Tests/CMakeLib/runcompilecommands [2-10->6/[email protected]] Linking CXX executable Tests/CMakeLib/testDebuggerNamedPipe
[1-10->7/[email protected]] Linking CXX executable Tests/CMakeLib/PseudoMemcheck/valgrind
[0-10->8/[email protected]] Linking CXX executable Tests/CMakeLib/PseudoMemcheck/purify
[0-9->9/[email protected]] Linking CXX executable Tests/CMakeLib/PseudoMemcheck/cuda-memcheck
[0-8->10/[email protected]] Linking CXX executable Tests/CMakeLib/PseudoMemcheck/memcheck_fail
[0-7->11/[email protected]] Linking CXX executable Tests/CMakeLib/PseudoMemcheck/BC
[0-6->12/[email protected]] Linking CXX executable bin/cpack
[0-5->13/[email protected]] Linking CXX executable bin/ctest
[0-4->14/[email protected]] Linking CXX executable bin/cmake
[0-3->15/[email protected]] Linking CXX executable bin/ccmake
[0-2->16/[email protected]] Linking CXX executable bin/ctresalloc
[0-1->17/[email protected]] Linking CXX executable Tests/CMakeLib/CMakeLibTests
It looks like newlines are being truncated somewhere.
|
I'll keep digging to see if I can find some other cause here. |
I find it a bit weird, that even in the old implementation, passing 10 will result in a string of size 9. Isn't this a bug? @evmar do you remember if there was a reason for that? I think d82e806 is the first commit which adds the 10 -> 9 unit test.
Maybe that should be fixed in the old implementation first.
Regarding this implementation: Using std::regex might be too slow, but it's a start and can be used as a reference implementation.