[
https://issues.apache.org/jira/browse/JENA-1188?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15316495#comment-15316495
]
A. Soroka commented on JENA-1188:
---------------------------------
Personally, I'd go for the fatal error, on the principle of least surprise.
Having all the graphs merged is not at all what I would expect, because i would
expect an export to produce something that I could reimport to get back to
where I began. Importing the merged data wouldn't give me anything like what I
started with. I hate to suggest making the CLI more complicated, but maybe a
new flag {{merge}}? So the logic would become:
1. --graph present and --merge absent, triple or quad formats permissible.
2. --graph present and --merge present, same as 1 (trivial case of merge just
one graph == just one graph).
3. --graph absent and --merge absent, only quads formats permissible, triple
formats throw a fatal error
4. --graph absent and --merge present, triple or quad formats permissible.
Triple formats behave as you describe above, quad formats merge into the
default graph
> sdbdump refuses to output a graph as Turtle
> -------------------------------------------
>
> Key: JENA-1188
> URL: https://issues.apache.org/jira/browse/JENA-1188
> Project: Apache Jena
> Issue Type: Bug
> Components: SDB
> Affects Versions: Jena 3.1.0
> Reporter: Osma Suominen
>
> I upgraded SDB to the 3.1.0 release version after running a snapshot build
> from early February. This broke a script that I use to periodically back up a
> SDB database:
> {noformat}
> sdbdump --sdb=$SDBROOT/sdb-yso.ttl \
> --graph=http://www.yso.fi/onto/yso/ \
> --out=TTL >$TTLFILE
> {noformat}
> This now gives the error
> {noformat}
> Not a 'quads' language (try 'N-Quads' or 'TriG')
> {noformat}
> The culprit seems to be this commit made on 8 Feb:
> https://github.com/apache/jena/commit/cb449523c83832ceb9cf525b4b1e1f382c49c16f
> I can see the point of requiring a quad-capable output format when dumping
> all graphs (though I'm not sure it should be a fatal error even in that
> case), but in my case, I asked for a single graph only. I think the check is
> overly strict.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
