Abraini's Blog » Seneca

Presentation 2:

abraini — Tue, 14 Apr 2009 04:41:39 +0000

Ubuntu Python26 Build
First I had to upgrade Ubuntu’s Python from 2.5.2 to 2.6.1. Downloaded and extracted the Gzipped tarball from http://www.python.org/download/releases/2.6.1/ and ran the ./configure file. After the configure finished compiled with the supplied make file. After the successful compile Python 2.6.1 was installed to usr/local/bin/python2.6 dir and the system was successfully upgraded. To compile OpenOffice.org on Ubuntu I checked out the latest Python26 child work space from the OOo SVN service. Followed the instructions on the OpenOffice.org Wiki website. Downloaded and installed the required packages and then used the same build flow as a Fedora build. After a short while I had a successful Ubuntu Build. With the build complete I installed the OpenOffice.org suit following the instructions provided by Ladan’s blogpost. I got the same error at the end but was able to run the suit from the install directory just as Ladan had.

Fedora Python26 Build
The Germany system was brought back online by C Tyler early April. Luckily the home dir’s were safe and this saved me some time not having to check out the cws again. And then I went and brought it down again. I had installed Java 6 over an install of Java 5 that Dave was using (major oops) and had caused conflicts between the two. C Tyler got back to me letting me know he and Dave had done some work to allow both versions of Java to run on the machine for me. This was much appreciated. I compiled Python 2.6.1 on Fedora as I did on Ubuntu but I did not install it over the primary Python install. I installed it to ~/bin/python2.6 and attempted to rebuild OOo on Fedora. I received the same error I had before the Germany refresh.

SyntaxError: invalid syntax Warning: 'with' will become a reserved keyword in Python 2.6 .

From what I understood searching for this error online it seems that I was trying to compile 2.5 code with the 2.6 compiler. I tried removing the solver dir and running make clean and recompiling. The same error occurred.

The next step I need a successful compile on a Fedora machine and extensive Quality Assurance testing.

* QA-1:
build the testtools module and switch to testtools/source/bridgetest/pyuno and run

dmake runtest

* QA-2:
install the office, start python from the program directory and type 

import uno

* QA-3:
Start the office, choose Tools/Macros/Run macro ..., click on "OpenOffice.org
Macros", you should then have a pythonSamples Library at the bottom of the list. 

That's it. In case you have any problems you can't solve, let us know either on
this list or within the above issue. Attach patches to the above issue.

I found a testtools on the old cvs and a wiki page for it. Downloaded/Checked out the old cvs on to Ubuntu with

cvs -d :pserver:anoncvs@anoncvs.services.openoffice.org:/cvs checkout udk/testtools

With the testtools downloaded I’m having trouble compiling them and running them.

Project Presentations

abraini — Fri, 03 Apr 2009 19:55:05 +0000

I believe everyone has had a chance to present their progress in their projects. I now have an understanding of what everyone is working on. Many of the tools they were using could help us all on our own projects. Daeseon showed us a walkthrough of the code he was working on and introduced us to Uno. Tiago explained how he was using a Virtual Machine to make an OpenSolaris box. It was interesting to see that even though both Solaris and StarOffice are run by SUN they were working on projects to make free versions. Or perhaps not, seeing as how there seemed to be little support by SUN for OpenOffice on OpenSolaris. Ladan demonstrated the problem that was in OOo’s Calc, how Microsoft dealt with the problem in Excel and what their plans were for Calc’s solution. Bart explained to us how they wanted to be able to copy master pages between Impress windows, and pointed out some functions in Impress that would help him achieve this. And lastly Jerry gave us a quick demo of the additional 3D slideshow transitions they want to implement in Impress and the process they were going to use to implement them.

I think I covered the presentations so far, I hope I haven’t left anyone out. These presentations really help show the scope of OpenOffice.org and its complexity. These projects aren’t going to be finished in the next two weeks but it is interesting to see how everyone is getting started. I’m sure it will also be interesting to see how everyone has progressed since there first project presentations.

1st OpenOffice.org Compile Attempt

abraini — Fri, 30 Jan 2009 17:13:03 +0000

I got OOo to configure on Germany with some help but soon found out I was going to be late for work (and was, oops) so had to postpone the compile. After getting home from work, eating dinner and watching a movie I remembered “ooh yeah I can compile OOo now” so I did.

So I start the compile at Fri Jan 30 00:30:56 EST 2009 and the next morning the log file is 12MB. Checking the log file it seems the compile halted at Fri Jan 30 01:33:54 EST 2009. This is the error I got.

rpmbuild: /home/arbraini/OOo/m39wc/solver/300/unxlngx6.pro/lib/libnss3.so: version `NSS_3.10′ not found (required by /usr/lib64/librpmio-4.6.so)
dmake: Error code 1, while making ‘../unxlngx6.pro/bin/noarch/fake-db-1.0-0.noarch.rpm’

ERROR: Error 65280 occurred while making /home/arbraini/OOo/m39wc/setup_native/scripts
dmake: Error code 1, while making ‘build_instsetoo_native’

Now it seems I’ll be late for class (oops). On the up side, it was compiling using all machines with distcc.

distcc and distccd setup

abraini — Mon, 26 Jan 2009 02:39:00 +0000

Added a simple and somewhat bare bones setup guide to the Seneca Distcc Wiki, with much of the info coming from Daeseon Moon’s Blog Post. We got Germany, Ireland and Australia setup as a distcc cluster. Running distcc on any of the three will have it compile on all three machines. This took a little work on figuring out which was the proper subnet mask for the cdot network. The right ip masking is now in the wiki.

For some fun I’ll work on finding comparative compile times of one machine vs two vs three and I’ll post the results as a comment on this blog.

Install Ubuntu & OpenOffice

abraini — Thu, 22 Jan 2009 15:31:11 +0000

I spent the last day installing Ubuntu. The Ubuntu install wasn’t the hard part, in fact it was the easiest install of any OS I’ve ever had the dis/pleasure of running. I didn’t even have to burn a CD. I was to lazy to go get one so I chose to install from a USB drive.

The only issue was in order to boot from the USB the boot order of the drives in the BIOS needed to be changed so the USB memory stick would be at the top. This caused and issue with GRUB. When the GRUB boot loaded installed it recognised the Windows XP OS as being on a hard drive in another order then when the USB is not booting. A simple fix to the GRUB menu items and dual booting was gravy. Installing OpenOffice was just as easy with Ubuntu’s nice and GUI’d Add Programs menu.

The lengthy issue was trying to get the Native install of Windows XP to load in VirtualBox or VMware. This took all day and in the end I was unsuccessful. It would actually have been faster for me to do a clean install of the OS and all the necessary applications but I was stuck on this problem and I wasn’t going to give up.

Next Task: Compile OpenOffice from source.

“Cathedral and Bazaar” thoughts

abraini — Fri, 16 Jan 2009 21:05:52 +0000

I finished reading Eric Raymon’s “Cathedral and Bazaar” paper two days before writing this and so have had some time to let the ideas inside the paper stew. The paper does a very good job of contrasting the two methods of developing projects, the corporate Cathedral style of a few specialized “sages” working hard to come to the finished project on their own before showing it the world. And the freesource Bazaar method of release early, release often with a transparency that allows many eyes to scan the project before anything has been finalized.

With these “many eyes” the bug or issue is going to be easily apparent to someone and perhaps the solution equally easily identifiable. This allows for a rapid development that could not happen in the Cathedral method. But there has to be someone steering the ship if you’ll excuse the mixed metaphors. Eric was in charge of his test project and in the end all design decisions were up to him. It will be interesting to see how this method works where there aren’t just many “eyes” but just as many “hands” on the code.