Is a good idea to use a GUID in name of files generated by users?

Question

I'm building a application (a CMS) where user can upload files like images.

My question is how to rename these files to save.

I think generate a GUID (System.GUID.NewGuid()) to save a file is the best way to go. I'm right or exist better approach in this case?

Note: An example of the GUID that is generated: 7c9e6679-944b-7425-40from-e07fc1f90ae7. In that case a image file will be: 7c9e6679-944b-7425-40from-e07fc1f90ae7.jpg

Update:

Users will not interact directly with the name of the file.

Answer 1

Yes. But probably a much more convenient scheme would use a hash-sum (say the MD5-sum) of the contents.

That way,

• the generation of the filename is repeatable (in case something goes wrong, data needs to be migrated to a different server, content is shared across different intallations etc).
• you'd automatically share duplicate uploads. Of course, then you'd need to track who owns the file (and not delete it untill the last usage is deleted)

Note An example of a typical md5sum is 5eb63bbbe01eeed093cb22bb8f5acdc3 (for ASCII/UTF8 "hello world")

Edit in response to the comments (about hash collisions): True enough, you might get hash collisions with very large sets of documents. In that case, it is most common to use the hash sum + the length of a file to identify the 'content blob'. So you'd do something like:

 http://cms.mysite.local/docs/123986/5e/b63bbbe/01eeed093cb22bb8f5acdc3.png

for a png of length ~ 123Kb