* nntpd will now disable ln_log()'s sending its logging to the
console (stderr usually), to prevent them from being seen by the
newsreader, which then gets confused.
* Usage fixes and usage info update
* Fix memory leaks
* Other bug fixes
For a full listing, please see the ChangeLog supplied with the source
code.
Dom:
* Fix bzip output to flush if last deflate fills buffer
Jody Goldberg:
* If the file is not an OLE2 file do not change the filepos.
* Handle broken ole files that specify codepage as signed short
* Fix reading ole files > 13.6 Meg on sparc/alpha (from Xavier Roche)
* Tune validation test for LPSTR properties to handle char width > 1
* Make the impl headers for the base interfaces public as requested
* Add some directory tree wrappers
* Work around a -fstrict-alias warning for SuSE
* Move to automake-1.7
* Fix gnome-vfs backend to open files with random access
* Support merging content from nested nodes for sax parsing
* Fix gtk-doc
Joseph Frazee:
* patch the spec file.
Morten Welinder:
* Properly chain finalizer.
* Fix gzip output to flush if last deflate fills buffer
Tor Lillqvist:
* packaging tools for win32.
changes:
Fixes from Max Neunhöffer (pyRXP) and peter Borocz (pdfmetrics).<br/>
ReportLab graphics now has some 3D charts.<br/>
Improved support for True Type Fonts in renderPM.<br/>
Added Henning von Bargen's LongTables optimisation.<br/>
Added Marc Stober's underline fixes.<br/>
Restored/improved platypus layout error detection.<br/>
New platypus FlexFigure flowable.<br/>
ReportLab is mostly jython compatible without Python 2.3 problems. <br/>
Pythonpoint improvements.<br/>
Many bugfixes.
the PERL defintionen in the Makefile resulting from Makefile.PL gets
set to 0 if LANG || LC_<something> is set to something else (read non english).
ok'd by jlam@
- The postdrop mail submission command could die with SIGHUP and
abort mail submission. This was observed with mail from cron jobs.
- The MySQL client aborted with complaints about multiple attempts
to register the same lookup table. This was observed in the proxymap
daemon.
- As a workaround for agressive SMTP command pipelining clients,
the Postfix SMTP server now allows SMTP clients to overshoot the
SMTP server recipient limit without triggering the server hard
error limit, as long as the number of excess recipients stays within
a hard-coded overshoot limit of 1000. If you have such clients then
you also need to specify "smtpd_error_sleep_time = 0" or else
performance will be poor.
- The LMTP client attempted to reuse a connection after timeout,
causing protocol synchronization errors.
- The trivial-rewrite server could core dump after temporary table
lookup failure. This was not observed in Postfix 2.0.
Georg Schwarz in PR pkg/25057:
- check for PKGSRC_COMPILER instead of USE_MIPSPRO:
USE_MIPSPRO would imply the former, but not the other way around.
That is, it's conceivable to have a system that has PKGSRC_COMPILER
set (to mipspro), but not USE_MIPSPRO.
- Allow passing of -Duse64bitint if ABI == 64 even if the compiler is not
mipspro. (Note: we could test this on a non-IRIX 64bit platform by
setting ABI to '64'. I think that should work, too, but I can't test
that right now.)
