Tracking issue for the functions for debug escaping char_escape_debug

[tbu-]

std::char::escape_debug, std::char::EscapeDebug.

[nikonthethird]

I hope this is the correct spot to mention that all examples of std::char::escape_debug use std::char::escape_default.

https://doc.rust-lang.org/std/primitive.char.html#method.escape_debug

[Mark-Simulacrum]

The above has been fixed; the function is correctly documented. Nominating for libs team discussion as to stabilization, I don't think we're going to get much information out of usage since this is such a niche API. I'm personally not certain I like the name: escape_debug seems odd to me, it's not really anything related to debugging I think.

cc @rust-lang/libs

[tbu-]

debug refers to the fact that it is used for the Debug implementation of String and &str.

I would prefer to call these functions just escape (best would be escape_default, but that one is taken). Could we maybe permanently deprecate escape_default in favor of something like escape_nonascii to make it clear that escaping all non-ASCII code points has nothing to do with "default"?

(I don't suggest ever removing escape_default, just deprecating it.)

[sfackler]

@rfcbot fcp merge

[rfcbot]

Team member @sfackler has proposed to merge this. The next step is review by the rest of the tagged teams:

• @BurntSushi
• @Kimundi
• @alexcrichton
• @aturon
• @brson
• @dtolnay
• @sfackler

Concerns:

• str resolved by Tracking issue for the functions for debug escaping char_escape_debug #35068 (comment)

Once these reviewers reach consensus, this will enter its final comment period. If you spot a major issue that hasn't been raised at any point in this process, please speak up!

See this document for info about what commands tagged team members can give me.

[BurntSushi]

Sorry, but I've lost context on this change and this issue doesn't appear to link to anything. What is the purpose of this new method? Is it just a rename?

[tbu-]

The function escapes characters the way the Debug implementation for str and char does.

[dtolnay]

If we are doing this, I would also like to have a plan for #27791 which includes <str>::escape_debug. Looks like there were concerns about consistency between char's and str's escape_default in #27791 (comment).

@rfcbot concern str

[brson]

Please change this sentence in the docs: "This will escape the characters similar to the Debug implementations of str or char." I assume this is not just similar, it is the implementation of 'Debug'.

[tbu-]

Currently. What if we wanted to change either of these but not the other? escape_default couldn't be changed because the documentation was too concrete.

[tbu-]

Maybe we could state that it does the same as the Debug implementation for str.

[ariasuni]

The only difference between escape_default and escape_debug seems to be that the first uses the range '\x20' ... '\x7e' to determine printable characters (i.e. characters which will not be escaped) where the second uses is_printable(). Why doesn’t it use the actual code for Debug then, instead of this gigantic char_private.rs which is used nowhere else?

[tbu-]

@ariasuni It's the other way around, the Debug implementation of char and str use escape_debug.

[ariasuni]

Oh, that explains a lot. So what does is_printable() is supposed to do?

[tbu-]

It's supposed to decide whether a codepoint is printed directly in the Debug output (e.g. if it's a character from some alphabet), or whether it should be escaped (e.g. if it is a control code point).

is_printable('a')
is_printable('α')
!is_printable('\n')
!is_printable('\0')

[ariasuni]

I get it, I’m wondering on what properties/logic this choice is based. Reading the (undocumented) source code is not helpful… I just understand that it’s based on the Unicode character class (also it probably should be updated for Unicode 10.0.0).

[tbu-]

See the original pull request: #34485.

[dtolnay]

@rfcbot resolved str

char::escape_debug is good to go. We will address the consistency question as part of #27791.

[rfcbot]

🔔 This is now entering its final comment period, as per the review above. 🔔

[leo60228]

I'm not sure of a better place to put this. Is this function guaranteed so that format!("\"{}\"", x.escape_debug()) returns valid Rust code that represents a value equal to x?