Error reporting

Using errno isn't great for error reporting, because most values are pretty useless. We often end up using EINVAL, and this doesn't allow callers to deduce a meaningful error message.

We should figure out a better way to report parsing errors.

mentioned in issue #3

The most of the cases we would be dealing with would be:

• invalid/NULL EDID
• invalid parameter requested
• parsing failed

I guess we can cover these with error messages or NULL/-ve return values.

Something which we really need is the ability to report string error messages for di-edid-decode purposes.

Also, it would be useful to be able to report non-fatal errors: unknown blocks which were skipped, garbage at the end of the EDID, etc.

mentioned in merge request !14 (merged)

Here is an example of edid-decode output, for reference:

Warnings:

Block 1, CTA-861 Extension Block:
  Display Product Serial Number is set, so the Serial Number in the Base EDID should be 0.
EDID:
  Base EDID: Some timings are out of range of the Monitor Ranges:
    Horizontal Freq: 28.125 - 352.801 kHz (Monitor: 255.000 - 255.000 kHz)

Failures:

Block 0, Base EDID:
  Basic Display Parameters & Features: sRGB is signaled, but the chromaticities do not match.
Block 1, CTA-861 Extension Block:
  Colorimetry Data Block: Reserved bits MD0-MD3 must be 0.
  Missing VCDB, needed for Set Selectable RGB Quantization to avoid interop issues.
Block 2, DisplayID Extension Block:
  DisplayID Base Block has no product type.
EDID:
  CTA-861: Native progressive timings are a mix of several resolutions.
  DisplayID: Missing DisplayID Product Identification Data Block.
  DisplayID: DisplayID expects at least one preferred timing.

EDID conformity: FAIL

All these come after the usual content printout?

Seem like we could do with something like

char *errors;
char *warnings;
bool conformant;

that could be queried from a di_info or a di_edid, and simply printed verbatim in di-edid-decode.

That's the "match edid-decode" side of error reporting. Not sure what compositor would need, but I can see Weston dumping these into its log.

mentioned in merge request !48 (merged)