On Thu, 1 Aug 2002, Troy A. Griffitts wrote: > >>[a-z,A-Z,0-9,_] are the only valid characters in module names. The > >>module name should be changed. > > > > > > Because? > > This standard is necessary for problems exactly as the one you've > experienced in the win32 frontend. These are internal names that need > to be usable in standard encoding mechanisms such as programming > languages, XML, HTML, etc. Please change your modules to adhere to > these restrictions and everything will work fine in the most possible > situations without extra work from others.
Ok.... So can you actually cite a standard you are adopting here? [a-zA-Z0-9_] is not sufficient. Using a standard of [source language id][separator][destination language id] is the only reasonable way I've thought of for naming glossaries in a consistent manner. That requires using [a-zA-Z0-9_\-] for the language ids plus one additional non-alphanumeric character to separate the ids. We can adopt XML Name, confined to Latin-1, if you would like. That adds [:\.] to the options but requires that the first character be in [a-zA-Z_:]. Alternately, we could add a safename() member to SWModule that takes the module name and returns a version where all characters outside of whatever you decide on are converted to characters within whatever you decide on. --Chris
