Display chained errors

[madig]

I'm trying to make norad cleanly chain errors instead of hoisting the error messages form lower into wrapping error messages. Compare what anyhow does:

Error: Failed to read glyph file from 'C:\...\MutatorSansLightWide.ufo\glyphs\A_.glif'

Caused by:
    0: Failed to read or parse XML file
    1: Expecting </array> found </dict>

vs. ufofmt's

[ERROR] norad read error: C:\...\MutatorSansLightWide.ufo: Failed to read glyph file from 'C:\...\MutatorSansLightWide.ufo\glyphs\A_.glif'

[chrissimpkins]

Excellent! That is really nice Nikolaus. Let me look into how to get at the data in the inner error messages. Please let me know if you have thoughts about how the approach modifications on the ufofmt side. I would like to support this.

[madig]

Hm, I think it is basically about recursively calling source() on the errors until you get None and print the messages of all, but maybe looking at the anyhow source code is more enlightening.

[chrissimpkins]

Is the example that yielded the error in your OP:

Error: Failed to read glyph file from 'C:\...\MutatorSansLightWide.ufo\glyphs\A_.glif'

Caused by:
    0: Failed to read or parse XML file
    1: Expecting </array> found </dict>

in the norad repo?

[chrissimpkins]

Note to self: https://docs.rs/anyhow/1.0.51/anyhow/struct.Error.html#display-representations

I don't have a problem with simply adding anyhow error handling in this project. It looks like we get the recursive calls handled with that library.

[madig]

I modified the UFO in question to end a <dict> with </array> somewhere, but the error messages in question should be in main.

[madig]

https://www.lpalmieri.com/posts/error-handling-rust/ contains an example on how to print error sources:

//! src/routes/subscriptions.rs
// [...]

fn error_chain_fmt(
    e: &impl std::error::Error,
    f: &mut std::fmt::Formatter<'_>,
) -> std::fmt::Result {
    writeln!(f, "{}\n", e)?;
    let mut current = e.source();
    while let Some(cause) = current {
        writeln!(f, "Caused by:\n\t{}", cause)?;
        current = cause.source();
    }
    Ok(())
}

[chrissimpkins]

Thanks Nikolaus!

Let's see if #44 addresses this. I'll prepare test sources to have a look at the output format from chained norad errors.

[madig]

Cool :) it might also be an opportunity to review the error hierarchy, as in, do the errors tell you what you need to know. See https://kazlauskas.me/entries/errors

[chrissimpkins]

Cool :) it might also be an opportunity to review the error hierarchy, as in, do the errors tell you what you need to know. See https://kazlauskas.me/entries/errors

sg!

[chrissimpkins]

Hmm. It looks like there was a big refactor in error types and now I can't build the branch in #44 based on norad master branch. norad::Error appears to be gone.

[chrissimpkins]

Sorted out in e4948b07ce088ad957c35fad9f125a402704094b

[chrissimpkins]

Looking good!

Here's what I see in the #44 branch if I mangle a cap A glif file:

[ERROR] norad read error: /Users/chris/Desktop/ufofmt-test/MutatorSansBoldCondensed.ufo
failed to load layer 'foreground' from '/Users/chris/Desktop/ufofmt-test/MutatorSansBoldCondensed.ufo/glyphs'

Caused by:
	failed to load glyph 'A' from '/Users/chris/Desktop/ufofmt-test/MutatorSansBoldCondensed.ufo/glyphs/A_.glif'
Caused by:
	failed to parse glyph data: missing close tag

and when I remove the glyph element that defines the glyph name:

[ERROR] norad read error: /Users/chris/Desktop/ufofmt-test/MutatorSansBoldCondensed.ufo
failed to load layer 'foreground' from '/Users/chris/Desktop/ufofmt-test/MutatorSansBoldCondensed.ufo/glyphs'

Caused by:
	failed to load glyph 'A' from '/Users/chris/Desktop/ufofmt-test/MutatorSansBoldCondensed.ufo/glyphs/A_.glif'
Caused by:
	failed to parse glyph data: wrong first XML element in glif file

This looks great! Nicely done @madig and @cmyr!