Re: Re: misleading message and exit code when "program too old"

[Johannes Waldmann]

Hi,

it's really not a big deal since I have a work-around
(ignore the exit code in my Makefile)

I understand that specifying and checking the version
can be useful. I am just saying that the actual message
and behaviour (exit code) is surprising.

>From the compilers I know (say, gcc)
there's a clear difference between:
* a warning (processing continues, exit code 0)
* an error (processing stops, no output, exit code != 0)

Where is the semantics of "\version" documented?
(It's not in the index?
http://www.lilypond.org/doc/v2.18/Documentation/notation/lilypond-index
- It appears in the PDF version
of the notation manual, but without explanation)

Is it something like the following?
"If lilypond version X reads "\version Y" in the input,
then if X < Y, it prints an "error" message and sets exit code to 1,
else it is silent. It will continue processing in both cases."

Why am I doing this - I have a bunch of 2.18 files that still
need to be converted to 2.19, that's why I am using 2.18.

In an ideal world, 2.19 would respect the "2.18"
specification in the file and behave accordingly ... but I guess
that would require a lot of work - basically keep all the
old/buggy behaviour in addition to the new/fixed one.

Since we don't have this, a 2.19 executable
should also warn if it reads a 2.18 file?
Because output could be different - that's the
same reason for the warning as in the other case.

That would give a lot of warnings.
Then they could be turn-on/offable ( -Wversion? )

- J.