On 28/06/2011 14:33, Konstantin Kolinko wrote: > 2011/6/28 Mark Thomas <ma...@apache.org>: >> >>> IMHO it is '_' that was treated as special char, because mangleChar() >>> prepends '_' as well. Otherwise the mangling will be unreversible. >> >> Does that matter? Do we need to able to reverse this? > > To avoid collisions between mangled names and names that did not need > to be mangled.
collisions are always going to be possible unless we go down the _jsp* route which - while spec compliant - would break even more stuff then than the original hello_world regression we were trying to solve. >>> The use case of '.' is just to use shorter name for a frequent use case. >>> E.g. "index.jsp" -> "index_jsp.class". >> >> It would be nice to keep that. > > Yes. It would be some surprise if names of (pre)compiled jsps will > change between minor versions. > >> I suppose we could add a >> periodToUnderscore parameter to makeJavaIdentifier() and essnetially >> have both approaches. The current one for file names, the new one for >> variable names. > > Or better just add a separate method. I never did like code duplication ;) I'm close to having a patch ready for this. Just need to run through the unit tests. Mark --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
