|include-sys-ioctl.patch||Include <sys/ioctl.h> even if <sys/filio.h> exists
We may need it for terminal ioctls such as TIOCGWINSZ. Fixes build on
|Colin Watson <email@example.com>||no||2014-08-29|
|spoiler-character.patch||Treat formfeeds as spoiler characters even not at start of line||Colin Watson <firstname.lastname@example.org>||no||2010-03-08|
|quoted-printable-segfault.patch||Don't segfault on poorly-encoded quoted-printable header lines||Colin Watson <email@example.com>||no||2010-03-08|
|subscription-prompt-overflow.patch||Prevent buffer overflow at the newsgroup subsciption prompt||Terran Melconian <firstname.lastname@example.org>||no||debian||2010-03-08|
|posix-conformance.patch||Conform to POSIX.1-2001 option changes||Colin Watson <email@example.com>||no||debian||2010-03-08|
|earlier-util-init.patch||Move util_init back up the initialisation order
doshell still doesn't work quite right if util_init hasn't been called.
This is noticeable when trying to use AUTHINFO GENERIC.
|Colin Watson <firstname.lastname@example.org>||no||2010-03-08|
|libtinfo.patch||Check for libtinfo separately from libcurses
Committed upstream (cvsps patchset 125).
|Colin Watson <email@example.com>||invalid||2011-10-19|
|thread-tree-stability.patch||thread tree stability (fixes broken tree navigation)
If two articles have the same date but different subjects and
different article numbers, and the article numbers compare in the
opposite order to the subjects, it can be impossible to navigate to
some places in trn4's thread tree. This is because the sibling
ordering of the articles in the tree is not stable, but depends on the
So it can be that going "up" from B to its earlier sibling A causes
(after rethreading etc.) A to be rethreaded after B, so now you would
go "up" from A to B. If there are even earlier siblings, they can be
unreachable (or at least difficult to reach).
It is IMO a bug that the thread tree capriciously rearranges itself
(at least, when all the articles are actually available etc. - there
are a lot of good reasons why rearranging the thread tree might be
necessary but it should be a convergent process and ideally should not
depend on the current article).
This patch fixes this particular rearrangement by extending the
partial order (on articles) implied by their dates into a complete
order by disambiguating on article number. This ensures that the
articles, when rethreaded, are linked in in the same order each time.
I have chosen a single-line formatting of the if's in the comparison
function because IMO that makes the regularity easier to see.
|Ian Jackson <firstname.lastname@example.org>||no||debian||2012-03-27|
|longer-line-buffer.patch||Segfault whilst applying killfile to article with long header line
This now reminds me of some investigations I did a few years back into a
similar issue in inn's nntpget. I found that a small number of clients
did not trim References headers when they grew past 1024 bytes, or perhaps
trimmed before appending the new message-id rather than afterwards.
Regardless, the number of articles with References headers > 1024
was non-zero, but few had them very much longer than that as most of
these lengthy subthreads involved at least one person with a non-broken
I therefore think the attached patch should be more than adequate.
Tested here and solves the problem on all the newsgroups I'm subscribed
to that have long enough threads to cause this issue.
|Nick Leverton <email@example.com>||no||debian||2012-08-24|
|extern-declarations.patch||Make more declarations in scan.h extern
Fixes builds with "gcc -fno-common".
|Colin Watson <firstname.lastname@example.org>||no||debian||2020-07-25|