Version: (using KDE KDE 3.0.99) Installed from: Mandrake RPMs Compiler: gcc 3.2 gcc 3.2 (Mandrake Linux 9.1 3.2-4mdk) OS: Linux Underscores in quoted-printable encoded attachments are translated in spaces. This is the right thing for quoted-printable encoded header fields (often seen in Subject: Headers with non-ASCII characters) but not for attachments. I have seen this (wrong) behaviour with HTML Mails send by an MS Outlook client. Earlier Versions did the Q-P conversion correctly.
Subject: Re: New: changes underscores in spaces in quoted-printable attachments A week ago I could reproduce this on one machine. But after a complete rebuild of qt-copy and KDE the problem is gone. I have no idea what could have caused this problem and as I can't reproduce it anymore I have to assume that the problem has been fixed somehow in the meantime.
Subject: Re: changes underscores in spaces in quoted-printable attachments On Tuesday 26 November 2002 23:31, you wrote: <snip> > Marc? Is it o.k. to commit this workaround? <snip> *beep* no patch *beep* :-D Also not in bugzilla. Ah, OK, found it. Is this causing a compile error or is it triggering a bug in gcc? What's the bug exctly? That KMail produces ' ' -> '_' in QP attachments? Or that it doesn't decode them? The mQEncding flag is set by the ctor and then never written again. I don't see how that could affect anything. A nice workaround is maybe to leave the const, but move the bool mQEncoding to before the bool : 1's and removing it's :1. maybe that's what causes gcc to miscompile? If that doesn't help, then go ahead with the patch as it stands. I will replace these conditionals with a variant of Strategy anyway (lookup table instead of class, though) when I get around to make the codec's text vs. binary aware. Marc
Subject: Re: changes underscores in spaces in quoted-printable attachments Am Wednesday 27 November 2002 00:49 schrieb Marc Mutz: > What's the bug exctly? That KMail produces ' ' -> '_' in QP attachments? > Or that it doesn't decode them? Underscores in QP message parts are decoded as spaces, so KMail produces '_' -> ' '. > The mQEncding flag is set by the ctor and then never written again. I > don't see how that could affect anything. The problem is that it is set true, although false is passed to the constructor. > A nice workaround is maybe to leave the const, but move the bool > mQEncoding to before the bool : 1's and removing it's :1. maybe that's > what causes gcc to miscompile? Leaving it const and moving it up at first place works for me, no need to remove it's :1. Patch attached. Regards, Wolfgang -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.7 (GNU/Linux) iD8DBQE95IJgGoxYmT/KkJkRAj3gAKCILm+F9ulpaPgjQPHx3pMiazjytQCdFYNa 978JDpAOr8G3zSIZgWAAeHE= =mSY3 -----END PGP SIGNATURE----- Created an attachment (id=536) kmime_codec_qp_v2.diff
*** Bug 51344 has been marked as a duplicate of this bug. ***
*** Bug 52328 has been marked as a duplicate of this bug. ***
This is still a problem for me with KDE 3.1 (rc7/final somewhere), even though the attached patch has been applied in CVS a while ago. I'm using Gentoo Linux 1.4, here are the relevant pieces of information from `emerge info`: Portage 2.0.46-r9 (default-x86-1.4, gcc-3.2.1, glibc-2.3.1-r3) ================================================================= System uname: 2.4.20-wolk4.0s-pre7 i686 AMD Athlon(tm) XP 2000+ COMPILER="gcc3" CHOST="i686-pc-linux-gnu" CFLAGS="-march=athlon-xp -mcpu=athlon-xp -O3 -pipe" CXXFLAGS="-march=athlon-xp -mcpu=athlon-xp -O3 -pipe" I believe this bug has caused a few unexplainably bad GnuPG signatures. Am I the only one who has this problem?