Patrice, As I tried to write before ... I think by far the best approach would be to s/\r?\n/\n/ on input files when they are read, and then have good NL-only source to deal with henceforth. Oh, and I guess s/\r/\n/ too, in case the input had CR without NL.
That is, remove all CR before NL and replace all other CR with NL, not worrying about whether a given file is "consistent" or not. I have seen plenty of input files with inconsistent EOL conventions, due to different people editing it and the tools/os's not doing the right thing. It's all just a needless hassle that we can get rid of, so we might as well do so, seems to me. What chomp does depends on the system that Perl is running on -- on Windows it will remove CRLF, on Unix just LF. It is useless for our purposes since the input file may have been created on a different system than the current one where makeinfo is running. I'm pretty sure that Perl does not do any kind of magic on input. Anyway, whatever magic Perl does or doesn't do, we don't need to worry about it or try to take advantage of it. Just remove/replace the stupid CR's and be done with it. wdyt? karl
