Jason An

I've found a rather extreme case of the edge swinging wide:  Here is the binary (weird arrow is near the bottom of `bench::with_get::h350330d79060ee3c:`): [bin.zip](https://github.com/Vector35/binaryninja-api/files/5021838/bin.zip)

I think it would take some benchmarking to see if it's better to do the 8 bit approach when byteOffset isn't a multiple of 4 vs constructing a new Uint8Array...

Python's `readline` module seems to do a lot of stuff on initialization (yuck) like rebinding tab to insert a literal tab, and ESC-TAB to use their completion function instead: https://github.com/python/cpython/blob/3ec719fabf936ea7a012a76445b860759155de86/Modules/readline.c#L1296-L1299

I think a hacky solution could just be adding `python 'import sys; sys.modules["readline"] = None'` (or some more intelligent mock than `None`) into the start of `.gdbinit` to just prevent...

I think the current setup-debug is pretty fragile (altering pacman.conf is pretty bad and also requires `geo.mirror.pkgbuild.com` to be the first mirror, which isn't configured by the script), so maybe...

IMO the message should be changed to "Searching for string", and if the string can be parsed as an integer, it should say "To search for the integer value instead,...

I think it'd be better if we had a setting to enable/disable automatically displaying the context for this case