Caolan McNamara

oocompare
oocompare currently a trivial pygtk script around gs -sDEVICE=png and netpbm tools to graphically diff postscript files.

msoffice 97 encryption
I was asked today why the ms word97 encryption hash is not a standard md5, but a modified one. I haven’t a clue of course, but I did read in levy’s crypto that when ms (and other companies) first tried to export crypto (back when the cat was still in the bag) that the NSA worked with them to find encryption that was acceptable to the NSA for export purposes, so it’s plausible that any tweaks were enforced on them from above to make it less powerful.

gcj and OOo
1.9.71s1 with integrated gcj1 workspace announced, should be possible to build ooo2 with gcj now. A few things cannot be built, but I ‘ll try and sort of few more of them out now.

Sitecom wl-113 usb wireless adapter, (usb ID 0df6:9071) is a zd1211 based thing, the original GPLed linux driver for the zd1211 is a .tar.gz inside this zip

A modified version updated to compile is here, and the patches for the Sitecom device ids is here. Theres a sourceforge project apparently being set up to maintain the driver.

I compiled up the modified one, and am amusing myself with wireless foo at the moment. my iwlist output confused me at the start as theres apparently at least two other wireless nodes in my apartment complex which being unexpected baffled me for a moment

systemboost is integrated, systempython is now ready for QA, dashboard updated.

build time for 1.9.71ms1 start 10:57:14 end 12:54:42 with ccache and nodep=true and without any –with-system flags.

fix for that ultra annoying rscpp “*** glibc detected *** double free or corruption” double fclose.

A few status updates on the personal wishlist for 2.0 dashboard (where I’ve made the lines straight). The infamous gnome filepicker has made it through to integration. Building with systemboost is nominated. gcj1 is still ready for qa while a fix for the libgcj regression is apparently in hand, and the PK11_GetCertFromPrivateKey symbol is available exported in mozilla 1.7.5.

> cd sw
> rm -rf unxlngi4.pro
> date && build >&! /tmp/build.log && date
Fri Jan 14 11:34:29 GMT 2005
Fri Jan 14 11:42:36 GMT 2005
i.e. time elapsed was 8 minutes 7 seconds
> rm -rf unxlngi4.pro
> setenv nodep true
> date && build >&! /tmp/build.log && date
Fri Jan 14 12:32:21 GMT 2005
Fri Jan 14 12:37:55 GMT 2005
i.e. time elapsed was 5 minutes 34 seconds

wedding bugfixes
So now I have a working 1.9 OOo under fedora I could fiddle with making some RSVPs last night, and ran into a .doc import layout problem in OOo 1.9.69, hence this fix.

gcj1
Installsets of the gcj1 workspace here. (sunjava is the same workspace built with sun java to hopefully verify no regressions for non gcj users.)

The following dirs could not be built with gcj, apparently for the listed reasons and are disabled when OOo is build with gcj. So anyone with an interest in furthering the gcjing of OOo can examine those to see what can be done (wizards is the most important IMO)

./scripting/examples/java/makefile.mk:
-> javax.swing.JTree.getLastSelectedPathComponent
./scripting/java/makefile.mk:
-> org.mozilla.javascript.EcmaError
./bean/native/unix/makefile.mk:
-> jawt.h
./filter/source/xsltfilter/makefile.mk:
-> StreamSource not found in the declaration of the local variable “stylesource”
./filter/source/xsltvalidate/makefile.mk:
-> javax.xml.parsers.DocumentBuilderFactory
./hsqldb/makefile.mk:
-> sun.security.action.GetPropertyAction
./canvas/source/java/makefile.mk:
-> sun.awt
./connectivity/com/sun/star/sdbcx/comp/hsqldb/makefile.mk:
-> org.hsqldb.lib.FileAccess
./sj2/stardiv/controller/makefile.mk:
-> com.sun.star.lib.sandbox.ResourceProxy
./sj2/stardiv/net/protocol/appletresource/makefile.mk:
-> sun.net
./sj2/stardiv/applet/makefile.mk:
-> com.sun.star.lib.sandbox.ResourceProxy
./sj2/stardiv/app/makefile.mk:
-> sun.net.ftp.FtpClient
./wizards/com/sun/star/wizards/web/export/makefile.mk:
-> com.sun.star.wizards.web.data.CGSession
./wizards/com/sun/star/wizards/web/data/makefile.mk:
-> javax.xml.parsers.ParserConfigurationException
./wizards/com/sun/star/wizards/web/makefile.mk:
-> com.sun.star.wizards.web.data.CGStyle
./wizards/com/sun/star/wizards/tutorial/creator/makefile.mk:
-> com.sun.star.wizards.tutorial.creator.frame.Framework
./wizards/com/sun/star/wizards/tutorial/creator/frame/objectFinder/makefile.mk:
-> javax.swing.JTable.setRowSelectionInterval
./wizards/com/sun/star/wizards/tutorial/creator/frame/makefile.mk:
-> com.sun.star.wizards.tutorial.creator.frame.objectFinder.XFramework
./wizards/com/sun/star/wizards/tutorial/executer/makefile.mk:
-> com.sun.star.wizards.tutorial.creator.frame.Framework
./qadevOOo/tests/java/ifc/document/makefile.mk:
-> java.awt.print.PrinterJob.lookupPrintServices
./qadevOOo/tests/java/mod/_sw/makefile.mk:
-> ifc.document._XFilter.FilterChecker
./qadevOOo/tests/java/mod/_sm/makefile.mk:
-> ifc.document._XFilter.FilterChecker
./qadevOOo/tests/java/mod/_xmloff/Chart/makefile.mk:
-> ifc.document._XFilter.FilterChecker
./qadevOOo/tests/java/mod/_xmloff/Draw/makefile.mk:
-> ifc.document._XFilter.FilterChecker
./qadevOOo/tests/java/mod/_xmloff/Impress/makefile.mk:
-> ifc.document._XFilter.FilterChecker
./qadevOOo/tests/java/mod/_sc/makefile.mk:
-> ifc.document._XFilter.FilterChecker
./qadevOOo/runner/util/compare/makefile.mk:
-> convwatch
./qadevOOo/runner/convwatch/makefile.mk:
-> javax.imageio.ImageIO.write
./rhino/makefile.mk:
-> javax.swing.JTextArea.replaceRange
./sandbox/com/sun/star/lib/sandbox/makefile.mk:
-> sun.applet.AppletAudioClip

Successfully built gcj1 with gcj, had to use rpm –install –nomd5 to install the core rpm created from the OOo buildtime rpm creation to get around libgcc_s.so.1 cpio MD5 sum mismatch. Odd.

gcj
gcj1 workspace for OOO 1.9.X looks good, managed to finally build an installset with it. (note head gcc4′s gcj/gij + normal gcc/g++ 34). Getting a failure at installtime though, doesn’t look related to gcj1, resyncing to SRC680_m69 and hopefully all will be well by w/e
status
A status dashboard diagram for the OOo work I’d like to seen in place for 2.0. Will try to keep updated.
house
The house, still incomplete. finished by end jan looks like a little white lie from the developers.