Progress this week
Achim’s download and indexer tool evolved to be a simple and easy tool to get the offline Wikipedia database files to yor device. Sorry to say but I wasn’t so productive this week. I’ve only fixed the bug that the “Home” folder on 1.1.3 was “/var/root” and not “/var/media”. This is necessary on 1.1.3 for a more or less one touch installation together with Achims tool and the installer.But because this was the only change I did I will not release a new version only because of this single change.We’ve did some changes to the Google Code page and the forums page. First, you’re not able to make comments on Google Code Wiki anymore. Please use the forum instead. On the forums I’ve change the “feature request” section to “Feature request and bugs”. So let’s add bug also to this section. In the past it was a bit anoying to look into three different place to find out what’s up.
März 9th, 2008 at 02:22
Will the GUI indexer extend to include downloading of images?
März 9th, 2008 at 21:30
It is already possible to upload the prepared images.bin or images_xx.bin with the tool. Simply use step 3 alone. Downloading and packing the images still has to be done with the ImageGetter.exe and pack.exe.
I hope that in the future well find a way to provide ready-to-install image packs somewhere (problem: they are huge), so there would be no need to include that procedure in the GUI…?
BTW, Sanford, in your installer package, do you remove the ~/Wikipedia hierarchy with the *.bin files on uninstall? I think that would be a good idea if that’s not already the case.
März 10th, 2008 at 02:07
Achim-
I did not remove the ~/Wikipedia hierarchy as I thought it is more “User” data but that can be added rather easily.
Also, while you are here, can you please take a look at the problem I reported in http://haukap.net/2008/03/03/a-gui-indexer-tool/ ? Thanks a lot in advance.
Tom-
I tried using your pack.exe for Windows (version 0.52) today and it seems it is not a constant-memory program. It is consuming like 260M memory when only 30k (out of 60k) images is packed. Can you look into the problem? (Or do you think it’s because of some stupid memory handling by Windows OS?)
März 10th, 2008 at 14:49
The pack.exe has a memory leak. That is already fixed, see this thread. http://wiki2touch.ipodhelp.de/viewtopic.php?id=29
Maybe we should keep all questions to the forum, I look there regularly, but not here at the blog.
Regarding your problem, could you post the logfile to the forum, please?
And removing the database during uninstall would be better, I guess. The average user who dowsn’t want the app any more probably doesn’t care about the databases himself, he just wants to uninstall…
März 10th, 2008 at 18:52
Sanford-
as Achim said, the pack.exe has a bug. Today I’ve written some code to convert characters in tradional Chinese to simplified Chinese. Hard because I cannot check by the character if it’s right or wrong.
Now I have a problem. I found at least once char in traditional Chinese (Unicode 0xE4, 0xB9, 0xBE) which converts to (Unicode 0xE5, 0xB9, 0xB2). But for the later there are two transition back. The first and the the second to (unicode =xE5, 0xB9, 0xB9).
So is it true tahn you can convert from traditional to simplified but not back and getting the same result?
I didn’t found any other char now, the remaining looking to be unique.
-Tom
März 10th, 2008 at 18:54
Achim, Sandford-
as long as the installer is not writing a file it should remove it. As long as we have no option to choose “remove data also” or so I would prefere to leave the file on the device.
IMHO the upload tool should be able to remove it because this tool has written the data. Only my two cents and there are good arguments for both ways.
-Tom
März 11th, 2008 at 16:53
Tom-
Thanks for looking into the Chinese translation problem. What you have noticed is indeed correct, sometimes a two traditional characters will have the same simplified form, take the example that you found:
E4B9BE is the traditional character 乾 meaning “dry”
E5B9B9 is the traditional character 幹 meaning “to act”
E5B9B2 is the simplified character 干 of both characters above (and is also a traditional character meaning “interrupt”)
The most ideal translation between the two forms of Chinese would be like this:
E4B9BE matches E4B9BE and E5B9B2
E5B9B9 matches E5B9B9 and E5B9B2
B5B9B2 matches E4B9BE, E5B9B9 and E5B9B2
However the usual way of implementing traditional / simplified chinese translation is just treat the two arrays as lowercase and uppercase characters. There is simply no end when you start to consider special cases because there are also variations on translations between multi-character words too.. and that will become way too complex. (Which is the case with the zh-wikipedia right now, in some pages there are even different “spellings” for a term in the different regions)
————–
About the installer package - I have made an experimental package that asks the user whether he wants to remove the data files as well - I will migrate this change once you release the next version of Wiki2Touch since you need to fix the /var/root thing for it to work anyway.
Achim-
I do not have the logfile as it simply refuses to run.