Hi,
i have seen that the code from tntzenoreader has been migrated to the openZIM subversion repository. Thank you Tommi for this work.
I have consequently adapted the Kiwix code and I'm now almost ready to release a beta of Kiwix 0.8 for Linux. I will do that next week and I will certainly also provide nightly builds in a near future.
But, now I need to build multiple ZIM files to "feed" it : I want in the next 2-3 months to do following: * Like you know, we are working on the last details on a 30.000 good and essential articles selection in English (issued from Wikipedia) * I want to do a ~50.000 big selection from essential articles issued from Wikipedia in French and in Spanish. * I want to make the Ubuntu documentation, issue from the Ubuntu-fr wiki (we have had some collaboration in the past, s. dokukiwix) * I want also make ZIM files with other selections like the SOS children village one.
All these ZIM files will be with pictures.
Unfortunately, almost every time I try to run the zimwriter on a "big" selection, it dies. This is not something new, we have talk about that during our meeting... but I write again today to ask Tommi if he had have time to have a look to this annoying issue... and otherwise maybe to try to motivate him to do it ;)
An other point essential point IMO, is the "Welcome page" header, I have seen that the new header was integrated, but I do not know how to specify it to the zimwriter.
Regards Emmanuel
Hi Emmanuel,
I have done some significant changes in libzim as well as updated the format of zim files. I have not yet checked in any code, since it is not ready yet.
The documentation about the file format is partly updated but not yet final.
The writer has got also some significant changes, so that it may be possible to create a zimwriter, which makes a zim file from a directory directly without using the database. All data is fetched via a interface class and collected in memory. I have calculated, that the memor usage is not that high, even when I keep the directory entries of all articles in memory.
I have had not that much time to work on the remaining issues, but I feel, that the new implementation is more elegant, flexible and robust. I hope to work on these issues soon. The target is to get it ready much before linux tag so that we have enough time to test and create the DVD.
Tommi