dj(1) null-byte padding: -A or -a ''? #101

New Issue

trinity · 2024-04-29T17:23:28-06:00

trinity commented

2024-04-29 17:23:28 -06:00

I wonder if the way to configure the padding byte used with synchronization of partial reads to '\0' should be changed from -A to -a ''; that is, -a with an empty but present option argument.

Changing this now won't make much of a difference for existing code; dj(1)'s padding and dd(1p)'s conv=sync are both very rarely used, the dj(1) feature only existing for parity with POSIX.

There's precedent for both uses in other tools.

It may make more conceptual sense that dj -a "\0" works, but it may make less conceptual sense (to a novice) that dj -a '' -i file works but dj -a -i file does not.

Both uses would be treated as special cases in the code.

I'm leaning towards -a '' because I think it makes more sense and removes some clutter from the command synopsis.

I wonder if the way to configure the padding byte used with synchronization of partial reads to `'\0'` should be changed from `-A` to `-a ''`; that is, `-a` with an empty but present option argument. Changing this now won't make much of a difference for existing code; dj(1)'s padding and dd(1p)'s `conv=sync` are both very rarely used, the dj(1) feature only existing for parity with POSIX. There's precedent for both uses in other tools. It may make more conceptual sense that `dj -a "\0"` works, but it may make less conceptual sense (to a novice) that `dj -a '' -i file` works but `dj -a -i file` does not. Both uses would be treated as special cases in the code. I'm leaning towards `-a ''` because I think it makes more sense and removes some clutter from the command synopsis.

trinity added the

question

label 2024-04-29 17:23:28 -06:00

trinity commented

2024-05-18 19:27:49 -06:00

It would not be too hard to make both usages work.