> > >> 2) Use some special `extensions translator' which automatically sets >> translators for files with known extensions. >
> Yeah, this might be possible. For a start, you could work on a filemux. > Then you need to couple the translator to start by filemux to the file type. I personally think having translators start automatically for a MIME like system would be extremely impressive. It allows you to 'cd' in to a file and see information in it, where otherwise you would have executed an application to interprete it's information for you. For example: gz, tar, zip, .. a mirriade of other compression formats html: a listing of all the other resources used inside the html file? xml: a virtual structure inside the file - this'd be great for debugging xml xslt: the same again. I think what I'm suggesting is that throwing different protocols at a file will have it react differently. For example, if you pipe into an xslt file an xml file, it should produce you the output of the translation. If you cd in to the xslt file you should be able to browse it. If you 'read' the xslt file you're editing it or viewing it with some other application. webpage.xslt < test.xml > output.xhtml view output.html/branding.jpg Again the same could be done for executables. If you enter in to it, you could be able to see symbolic links to the libraries it's going to use. Meta information about the program, etc. The list is endless. Not to mention the kinds of information available if you could have a node in /proc for every running process. There are many protocols that files could support. Piping, parameters, entering (like using 'cd'). >From what I've seen of Hurd so far, it does some of this. Ie; the 'Pager' protocol >when piping. Just some of my own ideas really. Michael Lucas-Smith _______________________________________________ Bug-hurd mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/bug-hurd