Hi,
until now, all import options for data in Marmotta were going through
the import web service, which is quite flexible and allow us to import
data from different scenarios (admin ui, client library, custom calls
with curl or whatever, and so on). But, of course, it has the inherited
overhead from HTTP. Therefore local import feature (MARMOTTA-145 [1])
was on my wish list for quite long time... And according my experiments
the performance is a big step forward.
Finally this week I've found some time to implement it, and it's already
available in the develop branch, and in the latest maven snapshots. On
the wiki you could find some more information how it works [2]; but
basically Marmotta is watching a local directory, where it imports all
the files copied there. I've already collected some interesting features
(target context names [3][4] and compressed files [5]) and bugs
(importing at boot time [6][7]).
Regarding one of this features (MARMOTTA-293: support fully qualified
context names importing locally), I'd like to ask the marmotta community
what option would be preferable for you. Basically in a coffee break I
discussed it with Jakob, and he suggests to make use of a config file
for such things; but personally I'd prefer to make it simpler, for
instance just using a url-encoded directory name for those cases. But
I'd like to listen other opinion before implement it; any other opinion?
Ideas would be welcomed ;-)
Cheers,
[1] https://issues.apache.org/jira/browse/MARMOTTA-145
[2]
http://wiki.apache.org/marmotta/ImportData#Import_data_via_the_local_directory
[3] https://issues.apache.org/jira/browse/MARMOTTA-291
[4] https://issues.apache.org/jira/browse/MARMOTTA-293
[5] https://issues.apache.org/jira/browse/MARMOTTA-296
[6] https://issues.apache.org/jira/browse/MARMOTTA-294
[7] https://issues.apache.org/jira/browse/MARMOTTA-295
--
Sergio Fernández
