gnaegi wrote:
do you want Apple to become the new MS where bugs must be accepted?
No. I want bugs to be fixed. There is only one route to getting bugs fixed, and it isn't in this forum.
Software that worked with Lion and Mountain Lion is broken with Mavericks. Our product which uses WebDAV code from Tomcat7 works flawlessly with Win XP, Vista, 7, 8 native WebDAV mount, even though you explained how difficult those clients can be. Curiously this time only the Mavericks client does not work.
I'm sorry, but that is not the way that software bugs work. Software must be written to published standards, in this case RFC 4918. These standards say nothing about client or server versions. They just specify what a client or server must do. Beyond that, both clients and servers are free to change their software in any way as long as it continues to comply with the standard. The greater they deviate from accepted practice, the more likely that there will be compatibility problems. Software that is (relatively) bug free will continue to work when either client and/or server code changes, as it always does. Sometimes, software works in spite of bugs and valid changes might expose them.
It is difficult to say where the bugs is, we spent a lot of time debugging and compared the http requests from Tomcat and Apache code with sniffers. But could not find and relevant difference. It could be in Apples code, it could be in the server code. Nobody really knows. Most software bugs can be solved on both sides, and apple has sure some special code in their client to work well with IIE and Apache. This is why we suspect it works with Apache, because Finder detects that this is an Apache server and then does something different than in other cases.
The only custom code in Apple's WebDAV software that I have seen is for IIS. It could be a problem with Apple's code, but I don't see it. I don't see how you can claim that software breaks with Tomcat but works with Apache and that you don't see any difference between the two. If there is no difference, then both work fine and the problem is now resolved.
It would be nice if they also tested with Tomcat7, this is all I ask for. Tomcat7 WebDAV implementation is also used by many people around there.
Why should Apple test with Tomcat? Why shouldn't Tomcat test with Mavericks? It seems Tomcat is the one doing something unusal with locking. I know a fair bit about WebDAV but had no idea that Tomcat had its own WebDAV server implementation. Tomcat is a Java environment for corporate IT, not anything that regular people would ever use.
Curios: does your great WebDAV server support WebDAV locking at all? Did you write from scratch or did yo use Apple's developer API to build the server? This might explain why your server works.
Yes. It fully supports the WebDAV protocol. That is required because the OS X WebDAV client is one of the few to require locking. Hmmm... Tomcat does funky things with locking and most clients don't support locking. OS X requires locking and has problems with Tomcat... Ah. Never mind. I'm sure there is no correlation there. 🙂
All software is based on APIs at some level. Tomcat uses the Java APIs. There are no APIs from Apple that provides assistance with writing WebDAV software, or even server software of any kind for that matter. While I did use Objective-C instead of Java, all of my WebDAV and HTTP server code is 100% my own. Only the lowest levels of networking operations use any Apple APIs. I just sat down with the RFC and implemented it. Locking with Mavericks works as expected. I do optimize for OS X's behaviour, but that is on a much higher level.
BTW: the reference to Tomcat you linked is from 2001 - boy are you kidding?? Do you know what OSX version was released in 2001 - it was 10.0! At that time few Mac users had ever heard of WebDAV.
I know nothing about Tomcat. Has it changed since then? Has the specific, very unusual method for handling WebDAV lock tokens changed since then? I don't use Tomcat so I don't care. I use Mavericks, Mountain Lion, Apache, and my own servers and they all work fine with each other. If this is an Apple bug, then it is the responsibility of Tomcat users to identify where it is broken, under what circumstances, and send a bug report to Apple. Apple may fix it and may not. Any fix will come later rather than sooner. If you, or your customers, depend on Tomcat to work with Mavericks, then you might want to consider exercising your open source rights and changing the Tomcat server yourself. Isn't that the whole point of open source? Why then are all of these open source users sitting on their hands and waiting for a proprietary software vendor to do something? It doesn't take a rocket scientist to realize that 3rd party corporate Java-based IT services are not big priorities for Apple.