>>Hi, >> >> >> >>Webfolders seems to have a nasty way of doing PUT requests : >> >> >>The context : >> >>Here is what you can see in webdav_access_log : >> >>"HEAD /b2web/jumal/private/INSTALL.TXT HTTP/1.1" 404 607 >>"PUT /b2web/jumal/private/INSTALL.TXT HTTP/1.1" 201 >>"HEAD /b2web/jumal/private/INSTALL.TXT HTTP/1.1" 200 - >>"PUT /b2web/jumal/private/INSTALL.TXT HTTP/1.1" 204 - >>"HEAD /b2web/jumal/private/NSTALL.TXT HTTP/1.1" 200 - >> >>Adding some debug code, I realized that webfolders were >>-> firstly doing a PUT request to create an empty file >> (contentLength was defined but equal to 0) >>-> only then performing a PUT request with the file's content. >> >>I'm pretty sure of this behaviour. The only thing I can't really >>understand is why second PUT returns 204 ? Are log writtings >>synchronized ? >> >> >>Implications : >> >>When we receive a PUT request that would exhaust user quota : >>-> the first PUT request is accepted (contentLength is ok) >>-> the second one is rejected (contentLength is too big) >> >> >>Result : >> >>All refused upload requests are ending by an empty file creation. >>And you can even lost a file that was having the same name but was >>ligther. >> > > waitaminute. if the first HEAD request doesn't return 404 but rather > 200/204 (i.e. a resource does already exist at the requested URI), > does web-folders STILL create an empty file ??
I've not checked contentLength headers but there are 2 PUT requests in this case too. Here is what you can see in webdav_access_log : "HEAD /b2web/jumal/private/INSTALL.TXT HTTP/1.1" 200 - "PROPFIND /b2web/jumal/private/INSTALL.TXT HTTP/1.1" "PUT /b2web/jumal/private/INSTALL.TXT HTTP/1.1" 204 - "HEAD /b2web/jumal/private/INSTALL.TXT HTTP/1.1" 200 - "PUT /b2web/jumal/private/INSTALL.TXT HTTP/1.1" 204 - "HEAD /b2web/jumal/private/INSTALL.TXT HTTP/1.1" 200 - > if yes, that would be very evil IMHO. imagine a webdav-server with > auto-versioning. you'd get one extra revision (the empty file) for > every actual change made. Effectivelly :-( I am using webfolders version that is included with IE5. I don't know if this behaviour has been changed later or with window$ XP JP > -chris > ________________________________________________________________ > cmlenz at gmx.de -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
