You probably already noticed it. I have no major problems with Jimmy Wales, but given that this spam header will not be hidden even after you donate (since I browse Wikipedia without logging in), this is really just spam. Here is a nice greasemonkey script to get rid of it.

The Document Foundation guys made an interview with me. In case you find out there something new about me, happy reading!

I just released 0.8.4. The happy thing about it is that basically the only fix I added to this release is to make it work when python is python3k (so we need to use python2.7 to get python2 support), but the rest is just merge from:

Now the total number of contributors increased to 7. :)

So I wrote a short script to find undocumented classes in the LibreOffice source code - I don’t mean when some of the public methods are not documented, but the case when the class itself doesn’t even have a one-liner mission statement.

Of course it should be used with care - if someone starts adding bullshit or misleading documentation using this script, that just makes the situation even worse.

The patch generated a nice thread where basically they argue if it makes sense to have a KDE-like API documentation for LibreOffice or it’s just better to refactor code to have self-documenting function names, etc.

At the end of the they, looks like there are people who find it great. :) (In the meantime that patch has been pushed, of course.)

Yesterday I wanted to print a PDF document: two pages per sheet, and of course I wanted duplex printing as well (on the short side).

Given that the hardware (the printer) was not capable of duplex printing, I needed some software workaround. If you want one page per sheet, this is trivial, as the print dialog of any PDF reader will let you select "odd pages", then you re-feed the printer with the output paper and you select "even pages" and that’s it.

OTOH, if you want 2 pages per sheet, then first you need some trick first to create two PDF from the original: the first containing the pages 1, 2, 5, 6, etc. - and the other containing 3, 4, 7, 8, etc.

After all this isn’t really hard using pdftk. First, the input PDF was around 150 pages, so I wanted to split the input to one file / page:

Then let’s move the pages to the subdirs a or b, based on the above pattern:

Finally concat the files from the a and the b dir to a single file:

That’s it!

I tried to search back when did I create Java bindings for libwpd - it seems this was the first commit.

First I kept back the source code to get some money for it (there were a customer who seemed to be willing to pay, but then disappeared), then I just abandoned the idea and pushed it to SourceForge.net to some CVS repo, (I even mentioned it in a blog post), which is finally now converted to git.

Long story short, a week ago finally Fridrich released the first version of the code, after about 2.75 years. :)

Background: OSL_TRACE() is an internal LibreOffice macro that is like printf(), but it’s only enabled in debug mode. Without it, you would add a debug printf during development:

then you would comment it out when the code started working,:

then remove the comment when it breaks again, etc.

In the past, if you built a LibreOffice module, then you executed:

inside a module, and in case you wanted debug symbols and such messages you built using:

which increased the debug level to 2 (from 0), where 1 is required for debug symbols, 2 for debug messages. (See the OOo wiki page.)

The recent update, why I’m writing this post: Now this has been changed, and debug=t gives you debug symbols only, but no extra debug output. While this is great in most cases, sometimes you still want the old behaviour to get the debug messages. In that case the solution (as pointed out in the mailing list post) is to use:

(Hopefully this post will be handy for who missed that post, like I did the first time.)

One may have noticed that right now my rejourn tree is empty. That’s not an accident, after Ram accepted my 15th patch, finally the site runs an unpatched version of rejourn - of course with a custom config and design.

Of course I still have a few ideas on what to improve, so this is not the end of the story, but still - I’m happy he was so open for changes and new features that we reached this post. :)

I usually just download the gpx track from my GPS and delete the track from device each time I arrive home and I was at some interesting place before. Last week this was not the case, I was at Gyúró two weeks ago, then I borrowed by device, without clearing the log from it.

So the task was easy: I had a timestamp and I had the full log from the device, and I want to split it: before and after a given timestamp. It turned out that I had some log from Oct 28th as well, so I wanted:

Luckily the gpsbabel documentation is quite clear in this area (see Example 4.8), I just needed the following commands:

Update: two more notes:

Once I was ready with this year’s GSoC work I started to submit it back to OpenOffice.org. Before that was completed, LibreOffice was launched, though I thought if I started working on submitting it to OpenOffice.org as well, I’ll finish it. About 6 weeks later they now integrated (merged) the vmiklos01 CWS (the branch containing my GSoC work) to DEV300, what will became OpenOffice.org 3.4 later.