Bitbake error must be prefixed with a unique ID, for each error, in the form "BBxyyyy" where "x" is a error severity indicator ("E", "W", "I" or "D" for Error, Warning, Info or Debug) and "yyyy" is a numeric ID starting with 0001. E.g. instead of: "Only one copy of bitbake should be run against a build directory" bitbake output should be "BBE0001: Only one copy of bitbake should be run against a build directory" Having unique textual IDs for errors make grepping through the source a lot easier, to allow for easy reference of code without having to hardcode locations. E.g. a web interface could display relevant sections of code based on the error unique id by grepping through the source.
This is a pretty invasive change which also has maintenance implications. At the very least this needs to be raised nd discussed with the community before we go any further with it. Alternatives such as the linenumber and file information should also be considered.
We're not planning to do this right now, most errors can be found comparitively easily as the strings alone are fairly unique.