After needing to find a small delimiter for my data format I started wondering if I could use 0x1E-0x1F?
They are part of the control codes so I thought they might do something weird?
https://en.wikipedia.org/wiki/C0_and_C1_control_codes#Field_separators
1E and 1F were actually originally intended to be used as record and unit separators, respectively, so that’s actually not a bad idea. The description for those fields in the article you linked even mentions that they’re suited for use as field delimiters.
Generally a bad idea to use in-band signalling like that. They won’t do anything weird but consider what happens if the actual data contains them.
consider what happens if the actual data contains them.
Then you’d escape them by using another character in front. But if their data format is ASCII text or is guaranteed not to have characters below ASCII 32 then using ASCII delimiters is fine.
But who escapes the escape characters?
It’s escape characters all the way down
You can use Unicode pictures: ␜ ␝ ␞ ␟
Use emoji as escape characters
Then you can just use a conmon delimiter like comma or semicolon or something. It’s better even as you’re less likely to have something that seems to work until your exotic delimiter pops up in the data.
Better yet, use a commonly used data format like csv or json and don’t build your own
Indeed. Escape characters add a lot of additional complexity, footguns, and performance penalties.
Depends on if you want your data format to be strict ascii. If you don’t care, then sure, why not?
depends on your format ? if the format is binary anyway or has binary blobs (ie it needs a program that is able to handle octets outside the printable range) and using those characters does not introduce any ambiguities with the format then go for it . ANSI and related control codes all start with
0x1B
