Hi,
What is the benefit of the database dumps being archived/distributed in xml format instead of sql format? Converting the xml to sql takes a long time for big wiki's and people seem to have problems with this step, so why isn't the sql format available for download instead of the xml format?
cheers, Jamie
mwDumper is essential also for anyone wiling to replicate a wiki locally for
any purpose. There are alternatives such as xml2SQL or importDump.php but mwDumper is the most efficient in terms of correctness and completeness or speed sometimes.
bilal ==
Verily, with hardship comes ease.
Hi!
Converting the xml to sql takes a long time for big wiki's and people seem to have problems with this step, so why isn't the sql format available for download instead of the xml format?
Our dumps are not 'sql dumps'. We assemble them from all the different parts (memcached, multiple database instances) - so it doesn't really make sense to output in some specific format, as one needs a decent loading routine to load that data into various kinds of stores.
Currently only widely available and supported marshalling format is XML, so hence we use it.
Do note, that if you use SAX-based conversion, conversion to SQL is way more efficient than database ability to load that data, especially if you attempt to maintain all the crossindexes on the fly too.
Domas
wikitech-l@lists.wikimedia.org