'server side' filesystems

I don't know if anyone has implemented this idea yet, but here goes. A webserver stores the the webpages as normal files. When you access a URL it sends you a copy of that file. But off course there is server side processing. Sometimes webserver don't store the file, but store a programme to generate that file. The webserver runs the programme and send you it's output. Wouldn't it be cool if there was a filesystem like that? Some files would store the contents of themselves as normal, but some files would be programmes to generate the files. When a programme/process opens that file, the file contents would be generated and then the output would be sent to the programme/process.