On Wed, Oct 27, 2004 at 06:56:35AM +0200, Christian Perrier wrote: > > % "Esperanto" > > lang_name > > "<U0045><U0073><U0070><U0065><U0072><U0061><U006E><U0074><U006F>" > > lang_ab "<U0065><U006F>" % "eo" > > -lang_lib "<U0065><U006F>" % "eo" > > +lang_lib "<U0065><U0070><U006F>" % "epo" > > > By the way, while writing a few new locales, I found it very hard > because I followed the above way to represent characters (with Unicode > Hex values).
Yes, this is really painful, I am pretty sure that someone already wrote an Emacs mode to handle this notation more gracefully. > As I found no dedicated tools for doing so, I wrote the locale with > "gucharmap" openedand editing the locale file with a simple text > editor....with thus a high risk of errors. > > Later, I found other locales examples with things like the following > : > > (value from fr_FR transposed to the notation I want to enhance) > > LC_MESSAGES > yesexpr "<^><[><o><O><y><Y><]><.><*>" > noexpr "<^><[><n><N><]><.><*>" > END LC_MESSAGES > > ...which is a bit more readable and less prone to errors in the case > of Latin languages at the minimum. AFAICT this notation was used years ago before Unicode was accepted. These are symbolic names which are defined in so called repertoiremaps. The Unicode notation got rid of repertoiremaps in order to prevent compatibility problems, so going back to this notation is a bad idea. Writing a tool to convert between this notation and Unicode notation is quite trivial, and I considered adding one to belocs-locales-data to help writing locale files, but it would only handle latin scripts, so I am reluctant to add a tool which discriminates against non-latin scripts even if it would be really useful to me ;) Denis
