Summary: | HTTP etag not properly updated after vCard updates, updates impossible | ||
---|---|---|---|
Product: | [Unmaintained] kresources | Reporter: | Helge Hess <helge.hess> |
Component: | groupdav | Assignee: | Reinhold Kainhofer <reinhold> |
Status: | RESOLVED UNMAINTAINED | ||
Severity: | normal | ||
Priority: | NOR | ||
Version First Reported In: | unspecified | ||
Target Milestone: | --- | ||
Platform: | Debian testing | ||
OS: | Linux | ||
Latest Commit: | Version Fixed In: | ||
Sentry Crash Report: |
Description
Helge Hess
2005-05-25 22:53:12 UTC
HTTP Log information: Once a write succeeds after the initial fetch of the vCards: ---snip--- PUT /zidestore/so/helge/public/Contacts/24710.vcf?sn=Duck HTTP/1.1 ... If-Match: 24710:1 BEGIN:VCARD ADR;TYPE=work:;;;Entenhausen;;123; ... X-KABCResourceGroupDAV-storagelocation:webdav://helge@move.in.skyrix.com:80 /zidestore/so/helge/public/Contacts/24710.vcf?sn=Duck END:VCARD May 25 20:04:06 ogo-zidestore-1.3 [16134]: <<0x087D565C[SxPerson]>>D etag '24710:1' matches: 24710:1 HTTP/1.0 200 Ok etag: 24710:3 content-length: 0 location: http://move.in.skyrix.com:80/zidestore/so/helge/public/Contacts/24710.vcf ---snap--- Notice that the version in the etag is bumped to "23710:3". ZideStore also deliviers the location though it didn't change => ZideStore bug. Another change: ---snip--- PUT /zidestore/so/helge/public/Contacts/24710.vcf?sn=Duck HTTP/1.1 ... Content-Type: text/x-vcard Host: move.in.skyrix.com If-Match: 24710:1 ... BEGIN:VCARD ADR;TYPE=work:;;sss\ndef;Entenhausen;;123; ... END:VCARD May 25 21:00:50 ogo-zidestore-1.3 [16285]: <<0x087D55FC[SxPerson]>>D etag '24710:3' does not match: ( "24710:1" ) HTTP/1.0 412 Precondition Failed content-type: text/html; charset="iso-8859-1" content-length: 187 <?xml version="1.0" encoding="ISO-8859-1"?> <html xmlns="http://www.w3.org/1999/xhtml"> <body><h3>An error occurred during object publishing</h3><p>Precondition Failed</p></body> </html> ---snap--- The client still has the old etag. The user doesn't get any kind of error notice, to him it looks like the change was successful (and indeed, the data is changed locally). (added a hack in OGo r1055 to disable etag-consistency for KDE 3.4) Absolutely, Currently there's hardly any error detection in place. Or rather, the user isn't told about the errors that are detected. NOTE: this bug is only partitially about error handling. The bigger problem is that the ETag is not updated after a change ("editing breaks because the etag is not properly updated after a PUT"). BTW: "ZideStore also deliviers the location though it didn't change => ZideStore bug" This has been fixed. Dear reporter, the resources handling got redone for KF5 and KF6 based PIM applications. If you still encounter the issues you had in the past, please report them to the current maintained and updated PIM applications you use Thanks a lot! |