Hi Dennis, > With respect to netcdf (at least the C version), it is the > case that these characters can appear > unescaped: _.@+- > > It should be noted however that dot in particular causes > problems for accessing remote datasets through DAP because > the dot character is used in DAP constraints to specify > fields inside DAP Sequences or Structures or Grids.
Out of interest, what happens if you do make an (open)dap request which involves an attribute name that contains 1 or more special characters (not just '.')? Does it fail? Is it just '.' which is the villain, or any netcdf special character? I guess I was (naively!) assuming that occurrences of variable/attribute names in, say, URLS would be escaped using the normal encoding rules, e.g. test.opendap.org/dap/data/nc/sst.mnmean.nc.gz?blah&blah&attname="cf.foob ar"&value="baz" Several of the examples at http://docs.opendap.org/index.php/QuickStart appear to employ a similar mechanism. > This means that whatever choice you make, you need to > enshrine it in a standard somewhere so that at least there is > a chance that people will avoid it. Yes, naturally that would be the intention - *if* a consensus could be reached. > Personally, I would think that a two character sequence is > least likely to be used by others, but two underscores is > probably not a good choice. I wasn't proposing that :-) > I would think something like @@ > ++ might be a better choice. Either of those would be a tough sell, IMHO! Though it's always good to get alternative suggestions/perspectives. Regards, Phil _______________________________________________ CF-metadata mailing list CF-metadata@cgd.ucar.edu http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata