wmc takes only one inputfile as argument (see BUGS). The inputfile normally has extension .mc. The messages are read from standard input if no inputfile is given. If the outputfile is not specified with -o, then wmc will write the output to inputfile.{rc,h}. The outputfile is named wmc.tab.{rc,h} if no inputfile was given.

-B x

Set output byte-order x={n[ative], l[ittle], b[ig]}. Default is n[ative].

-c

Set 'custom-bit' in message-code values.

-d

NON-FUNCTIONAL; Use decimal values in output

-D

Set debug flag. This results is a parser trace and a lot of extra messages.

-h

Print an informative usage message.

-H file

Write headerfile to file. Default is inputfile.h.

-i

Inline messagetable(s). This option skips the generation of all .bin files and writes all output into the .rc file. This encoding is parsable with wrc(1).

-o file

Output to file. Default is inputfile.rc.

-u

Assume that the inputfile is in unicode.

-U

Write resource output in unicode formatted messagetable(s).

-v

Show all supported codepages and languages.

-V

Print version end exit.

-W

Enable pedantic warnings.

A new keyword Codepages is introduced to set both input and output codepages to anything one wants for each language. The syntax is similar to the other constructs:

Codepages '=' '(' language '=' cpin ':' cpout ... ')'

The language is the numerical language-ID or the alias set with LanguageNames. The input-codepage cpin and output-codepage cpout are the numerical codepage-IDs. There can be multiple mapping within the definition and the definition may occur more than once.

Unicode detection of the input is suboptimal, to say the least. It should recognize byte-order-marks (BOM) and decide what to do.

Decimal output is completely lacking. Don't know whether it should be implemented because it is a, well, non-informative format change. It is recognized on the commandline for some form of compatibility.