Archive for October, 2004

OOo 1.9.58/1.9.59

Thursday, October 28th, 2004

1.9.58
With these patches 1.9.58 can be built with…

./configure --enable-libart --enable-libsn --enable-cups --enable-fontconfig
--without-fonts --with-system-zlib --with-system-libxml --with-system-jpeg
--with-system-expat --with-system-freetype --with-system-curl
--with-system-sablot --with-system-nas --with-system-mozilla

The config_office test should correctly test to see if your mozilla headers and libraries are sufficient to build with –with-system-mozilla (i.e. include https://bugzilla.mozilla.org/show_bug.cgi?id=262274 fixes and ldap headers), rather than tell you 10 hours down the line :-)

Additionally a fix for an installation problem (when using the system-expat backend) is included in the patchset.

Patches naming scheme encodes if the patch is pending in a workspace or not, and if not if there is an open OOo issue for it.

configure2
nominated and apparently integrated. brings –with-system-libxml up-to-date again

writerperfect2
I resynced it to 1.9.59 and fixed the blocking merge conflict in sw/inc/iodetect.cxx (yes a .cxx in a header file, it’s a long story)

ineturl1/gtkfpicker2
Resynced to 1.9.58, and currently building for another tilt at making universally working installsets

autowidth/autoheight

Thursday, October 28th, 2004

A patch for the ww8 OOo filter. I can count on one hand the amount of patches for winword feature import/export that I’ve received in years. jiangdongjin is the man.

OOo and gdb/valgrind

Wednesday, October 13th, 2004

I’ve updated the gdb testcases for OOo/c++ debugging at http://people.redhat.com/caolanm/mygdbproblems.tar.gz. Now with gdb in FC3 only one of the tests fails. I’m not convinced that breakpoints in ctor’s are fixed, but it works in the test case, so perhaps it is :-) . Deferred breakpoints on class member names works, as does repeated execution of a debugged process when deferred breakpoints have been resolved in a previous pass.

MHU has taken on a proper fix for the commandline parsing issue that confuses OOo when run from inside [val|call]grind which I was hacking around with http://people.redhat.com/caolanm/callgrind/

Meanwhile, gtkfpicker2 and ineturl1 languish in Ready for QA.

1.9.55

Tuesday, October 5th, 2004

OOo 1.9.55:
With 1.9.55, these patches, and the following configure line (8 line building 1.9.55 howto)

./configure --enable-libart --enable-libsn --enable-cups --enable-fontconfig --without-fonts
--with-system-zlib --with-system-libxml --with-system-jpeg --with-system-mozilla
--with-system-expat --with-system-freetype --with-system-curl --with-system-sablot
--with-system-nas

then 1.9.55 builds against system components in a more friendly way. Only thing that is problematic is the ongoing requirement to patch mozilla to get xmlsecurity to link against it, see #moz262274#. But at least its down to just one symbol (PK11_GetCertFromPrivateKey), given this patchset, now. Drop –with-system-mozilla if you don’t want to patch your mozilla :-)

Sorting out system berkleydb is on hold while theres work under way to move from db3 to db4. Similiarly neon is out of date, and OOo needs to be updated to latest api, so –with-system-neon probably won’t work for now. I haven’t checked out system-python while python is also being updated to latest and greatest.

Additionally before building, setting

setenv ENABLE_GNOMEVFS TRUE

and

setenv ENABLE_GTK TRUE

will build the gnomevfs and gtk optional extras. These aren’t set from ./configure yet.

Seems to me that nas should have a –disable-nas seeing as I really don’t think its in real world use anymore. Perhaps a esd implementation could be stuffed in instead

workspaces:
The configure1.* patches are included in the configure1 workspace awaiting integration. There is no system-mozilla one until its clear what the PK11_GetCertFromPrivateKey story will be. The apparently gigantic ineturl1 isn’t as big as it looks, it implements the change from UniString to rtl::OUString[Buffer] for INetURLObject that I mentioned at the ooocon as a plausible performance improvement. A big self-extracting linux installer for ineturl1 is available here. Meanwhile gtkfpicker2 languishes in the wilderness until OOo Headquarters can build and test with gtk2.4 (which is soon I believe)

other:
Been trundling around on shiny new dublin tram system for the last couple of days sorting out my provisional driving licence. And have now successfuly negotiated getting my hands on one, fear me.