asdf info broken on Roman L2 files

[embray]

These seem to hit some limitations of and break the asdf info command tree display (running it with --no-tree --blocks works fine and just displays the block info).

There are at least a few problems:

• nodes with very long scalar (string) values

  • Some of these just break the display due to terminal line wrapping, even on a fairly wide terminal window
  • Some of them contain multi-line values which also completely break up the tree display
  • In either case these should probably be truncated to some maximum preview display width...

• Other cases where the tree limbs just get broken for no apparent reason -- in some cases even very deeply-nested structures are displayed perfectly fine, but sometimes there is random breakage like:

│ │ │ │ │ │ │ │ │ │ │   │   │ │     ├─[0] (scalar): -1
│ │ │ │ │ │ │ │ │ │ │   │   │ │     └─[1] (scalar): 1
│ │ │ │ │ │ │ │ │ │ │ │     ├─inputs (sequence)
│ │ │ │ │ │ │ │ │ │ │ │       ├─[0] (scalar): x0
│ │ │ │ │ │ │ │ │ │ │ │       ├─[1] (scalar): y0
│ │ │ │ │ │ │ │ │ │ │ │       ├─[2] (scalar): x1
│ │ │ │ │ │ │ │ │ │ │ │       └─[3] (scalar): y1
│ │ │ │ │ │ │ │ │ │ │ │     └─outputs (sequence)
│ │ │ │ │ │ │ │ │ │ │ │       ├─[0] (scalar): z0
│ │ │ │ │ │ │ │ │ │ │ │       └─[1] (scalar): z1
│ │ │ │ │ │ │ │ │ │ │ │ ├─inputs (sequence)
│ │ │ │ │ │ │ │ │ │ │ │   ├─[0] (scalar): x0
│ │ │ │ │ │ │ │ │ │ │ │   └─[1] (scalar): x1
│ │ │ │ │ │ │ │ │ │ │ │ ├─inverse (tag:stsci.edu:asdf/transform/compose-1.3.0)
│ │ │ │ │ │ │ │ │ │ │ │   ├─forward (sequence)
│ │ │ │ │ │ │ │ │ │ │ │   │ ├─[0] (tag:stsci.edu:asdf/transform/remap_axes-1.4.0)

Not really sure why...if there's some corner case not handled or what.

• But worst of all it crashes finally with:

free(): double free detected in tcache 2
[1]    2720801 IOT instruction (core dumped)  ../.libasdf/bin/asdf info ../tests/fixtures/roman_l2_wcs.asdf

[embray]

I also noticed somewhere in the output (though not even right before crashing) something like:

│ │ │ │ │ │ │               └─[1] (tag:stsci.edu:asdf/transform/compose-1.3.0)
│ │ │ │ │ │ │                 ├─forward (sequence)
│ │ │ │ │ │ │                 │ ├─[0] (tag:stsci.edu:asdf/transform/remap_axes-1.4.0)
│ │ │ │ │ │ │                 │ │ ├─M` (sequence)
│ │ │ │ │ │ │                 │ │ │ ├─[0] (scalar): x0
│ │ │ │ │ │ │                 │ │ │ └─[1] (scalar): x1
│ │ │ │ │ │ │                 │ │ ├─mapping (sequence)
│ │ │ │ │ │ │                 │ │ │ ├─[0] (scalar): 0
│ │ │ │ │ │ │                 │ │ │ ├─[1] (scalar): 0
│ │ │ │ │ │ │                 │ │ │ └─[2] (scalar): 1
│ │ │ │ │ │ │                 │ │ ├─name (scalar): xtyt2xyz
│ │ │ │ │ │ │                 │ │ └─outputs (sequence)
│ │ │ │ │ │ │                 │ │   ├─[0] (scalar): x0
│ │ │ │ │ │ │                 │ │   ├─[1] (scalar): x1
│ │ │ │ │ │ │                 │ │   └─[2] (scalar): x2

where that `M`` is clearly garbage. So there is likely some subtle memory corruption happening which could be responsible for many of these problems. Could even be a bug in libfyaml...