By limiting shortened buffers to those actually using tracking.el or
lui, we can gain a substantial performance improvement for tracking
said buffers (when shortening).
Alternatively, we can add a variable to only shorten currently tracked
buffers (ie: tracking-buffers), which will be even faster.
I actually prefer shortening only the currently tracked buffers even if it
leads to non-deterministic shortening. Do you like that solution more?
As you can see, we still have a way to go on this function (and others in the critical path), but it's a start. Unfortunately, it all seems to be in the shortening library (on this particular call).
By limiting shortened buffers to those actually using tracking.el or lui, we can gain a substantial performance improvement for tracking said buffers (when shortening).
Alternatively, we can add a variable to only shorten currently tracked buffers (ie: tracking-buffers), which will be even faster.
I actually prefer shortening only the currently tracked buffers even if it leads to non-deterministic shortening. Do you like that solution more?
See: #286, #346, #344
For me:
As you can see, we still have a way to go on this function (and others in the critical path), but it's a start. Unfortunately, it all seems to be in the shortening library (on this particular call).