Therefore we do not rely much on the readerwin member vars which caused
problems with parallel downloads. This should hopefully fix the remaining load-on-demand
problems. It definitely fixes#92739.
I only noticed that the OpenWith action takes ages but this is probably not
a result of the change.
BUG: 92739
svn path=/trunk/kdepim/; revision=380928
Make isFirstTextPart() return true if the node is the first text part of the message this node is a body part of. This fixes showing attached messages embedded with smart attachment view mode.
<till> mahoutsukai: Making Dirk happy is our prime objective.
svn path=/branches/KDE_3_3_BRANCH/kdepim/; revision=349380
Move the static next() implementation from attachmentcollector.cpp to be a public partNode method. Implements iteration in "document order".
Use both to make AttachmentCollector skip the first text part, which would otherwise lead to the reply text being attached to a reply :)
svn path=/trunk/kdepim/; revision=318219
Add isHeuristicalAttachment(), which returns true if either isAttachment() is true, or if either the name or filename parameters is non-empty.
svn path=/trunk/kdepim/; revision=318179
Small cleanup. Get rid of partNode::CryptoType, which wasn't used for anything. Determine the CryptPlug to use for mp/signed by looking at the protocol parameter instead of searching for app/{pgp,pkcs7}-signature children.
(without the CryptPlugFactory)
--and--
candidate_for_3_2_1: Don't interpret application/pkcs7-mime; smime-type=certs-only as a signed message, but show an icon instead (until we have integrated enough of gpgme to be able to show a nice summary of the import as inline HTML).
svn path=/trunk/kdepim/; revision=281273
Refactor partNode creation from a KMMessage into a static method on partNode and port users to it. In the process of this, I found out that KMMessage, KMSaveAttachmentCommand and KMReaderWin all used very slightly different methods to instantiate the root item. I think most of them were wrong:
KMMessage's way (originally spread over two methods...) leaks mainBody...
KMReaderWins' way adds firstBodyPart as a child of itself...
KMSaveAttachmentsCommand doesn't adjust the mimetypes.
So that's why this is experimental: I've combined the various ways with majority vote: two adjusted the mimetypes, so mimetypes are adjusted. Only one added a part as child of it's own, so that's no longer being done, etc.
--and--
Replace broken, complicated logic with clean and simple one. Parsing is cheap in mimelib. That gives us the full MIME headers also in the top-level container partNode, which is needed in ae_branch to get at the protocol-Parameter of mp/signed parts...
svn path=/trunk/kdepim/; revision=281270
Crash was caused by the following header
Content-disposition: attachment; filename=
when kmail tried to access s[0] where s is the empty string behind filename=. The obvious fix is to test for s.isEmpty().
CCMAIL: 47856-fixed@bugs.kde.org
svn path=/trunk/kdenetwork/kmail/; revision=187212
debug output (only for the moment) to the IMAP methods to find out why it's broken
(finally got a few IMAP accounts, hopefully today few fixes for it will land)
svn path=/trunk/kdenetwork/kmail/; revision=169280
Thiago Macieira
Laurent Montel
Reinhold Kainhofer
Carsten Burghardt
Marc Mutz
Ingo Klcker
Till Adam
Dirk Mueller
Daniel Molkentin
Stephan Kulow
Karl-Heinz Zimmer
Michael Haeckel
Zack Rusin
Till Krech
Carsten Pfeiffer