On Mon, Aug 17, 2009 at 3:12 PM, Mark Janssen <[email protected]> wrote:
> Hmmm for [] VMS comes to mind. > So fossil takes the common denominator for file names on all > platforms? It sounds that way. > Although that's certainly good for portability of the > repos, it does limit the artifact names you can use. Renaming the > artifacts to a name which is allowed, is not always an option. Perhaps > a warning would be more appropriate? It's a tough philosophical question for fossil. What is it to do if it tries to extract an artifact and cannot? It cannot know (without guessing, based on the combination of current platform and underlying filesystem type) that the problem is related to invalid characters. And even if it does determine that the filename is invalid, what could it possibly do? It cannot extract the file and it cannot rename an artifact (they're immutable). i don't see any way for fossil to do this, other than blatantly disallowing any questionable the characters. Imagine that you've checked in a file named "DoesThisWork?", and that works fine on your *nix box. Now someone with Windows tries to check it out and cannot get that file because his OS won't let a file with "?" be opened/created. Now the data is locked into fossil and cannot be taken out without: a) using a platform which supports "?" b) downloading the file individually from the repo web interface (easy if it's a text file, not so easy for binary). The ZIP file solution is not helpful here unless your ZIP browser also allows you to rename files inside the zip or rename them upon extraction. While i agree that it's sub-optimal, i don't see any other realistic behaviour for fossil here. -- ----- stephan beal http://wanderinghorse.net/home/stephan/
_______________________________________________ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
