single backslash \ at end of line could be disallowed within $'...' (and maybe others)

[bar-g]

Is it ok to pass the backslash (consumed by command line parsing?) on to the string literal, and it becoming a literal \ even though parse_backslash is off?

ysh ysh-0.19.0$ echo $'a\
> b'
a\
b
ysh ysh-0.19.0$

And this is a non-identical variation that is printing the same:

ysh ysh-0.19.0$ echo $'a\\
> b'
a\
b
ysh ysh-0.19.0$

[bar-g]

It seems to be adding a backslash:

ysh ysh-0.20.0$ $'a\
> b'
  $'a\
  ^~
[ interactive ]:10: 'a\\\nb' not found
#                     ^^ why two?

Similar to here:

ysh ysh-0.20.0$ a'\'b
  a'\'b
  ^
[ interactive ]:9: 'a\\b' not found
#                    ^^ why two?

[bar-g]

[ interactive ]:9: 'a\\b' not found may this actually rather mean:
[ interactive ]:9: $'a\\b' not found, i.e. having searched for r'a\b'?
(could make some sense as the prompt is saying ...$)

For the first example, I guess I would have expected it to fail with "Invalid char escape" (or print ab as with double-quotes)

The problem seems to be there is never an "invalid char escape" error, nor an escaped (dropped) newline (sameas in double-quoted strings) with no amount of backslashes before a newline.

[andychu]

It seems like this is identical to bash behavior

$ echo $'a\
> b'
a\
b


$ bin/ysh
ysh ysh-0.20.0$ echo $'a\
> b'
a\
b

The issue is that all strings in shell are MULTILINE strings. Single quotes, double quotes, etc.

\ at the end of the line is NOT an escaped newline -- that's only true outside quotes

[bar-g]

What you say does not explain why the single \ is parsed with parse_backslash off.

It's also against what happens within double-quotes. There the newline (and the backslash before it) gets consumed, and no "Invalid char escape" is necessary:

ysh ysh-0.20.0$ echo "a\
> b"
ab
ysh ysh-0.20.0$

Consuming is consistent with shell:

$ string="
        one \
        two \
        three \
       "
$ echo "$string"

        one         two         three 

Plus python seems parsing single backslashes in singe- and double-quotes just fine, and there is still a multifold of ways remaining to denote the same string using \b.

This makes me ask how well backed up and necessary is the breaking behavior?

And maybe it does make sense, to rather improve uniqueness of denoting strings with a strict_ option forbidding singelton-backslash-pairs -- instead of singleton-backslashes -- to maintain compatibility with shell and python.

[andychu]

Oh I see! Yeah actually that is a good point.

We want backslashes to always be doubled so there is one way of writing it

I'm not sure why it is the way it is

[andychu]

This is fixed - we now consistently disallow the single backslash then newline in YSH (details in commit description)

Will be out with the next release

Great testing, thank you!

[andychu]

Released! https://www.oilshell.org/blog/2024/03/release-0.21.0.html