Jeff-
I've got a test verifying a bug in FCRepo 3.6 (and earlier) for content
posted in the request body- I'll push a fix today (I think). There are two
places in the REST library in which signed ints are being used got
content-length in the kind of upload you have:
https://github.com/fcrepo/fcrepo/blob/master/fcrepo-server/src/main/java/org/fcrepo/server/rest/RestUtil.java#L74
(where it's implicit in the j2ee interface, so we just have to ignore it
and parse the header)
and at
https://github.com/fcrepo/fcrepo/blob/master/fcrepo-server/src/main/java/org/fcrepo/server/rest/RestUtil.java#L63
(our fault- that was clearly a place we missed when an earlier bug in this
line was fixed 3 years ago).
The latter bug causes a mis-reported length for the DS (0), whereas the
first causes your null-pointer. Did you file a JIRA issue for this?
- Ben
On Tue, Jun 18, 2013 at 2:43 PM, Jeff Minelli <[email protected]> wrote:
> Offline Eddie also suggested using fedora/upload. Problem is (or I
> believe it to be a problem) that hydra uses rubydora, and if I am correct
> (I have no doubts I could be wrong here, I did a cursory check of the gem),
> rubydora uses the same datastream REST method. If so this will cause some
> pain hydra side..
>
> If I ever get some spare time, I will attempt tests via upload and on 3.6+
> for completeness.
>
> Thanks,
>
> -jeff
>
>
> On Jun 14, 2013, at 3:38 PM, Benjamin Armintor <[email protected]> wrote:
>
> > Jeff-
> >
> > Embarrassingly, I'm still trying to put together an optional test in
> the fcrepo 3 test suite that tries your request. I'll get back to you
> about that, but in the meantime it appears to be the case that POSTing to
> fedora/upload works, so you could use the location that's returned as the
> dsLocation in a two-step process.
> >
> > - Ben
> >
> >
> > On Thu, Jun 13, 2013 at 10:47 AM, Jeff Minelli <[email protected]> wrote:
> > Output attached for a second test:
> > curl -H "Content-type:application/octet-stream" -vv -X POST -u
> "myname:mypassword" --data-binary "@3g.data"
> http://ss1test:8080/SStest2Fedora/objects/scholarsphere:5425k9692/datastreams/blah?controlGroup=M
> >
> > It fails in a similar way.
> >
> > -jeff
> >
> > On Jun 13, 2013, at 10:25 AM, Benjamin Armintor <[email protected]>
> wrote:
> >
> > > Jeff-
> > >
> > > Are you doing a multipart upload? If yes, can you try posting the
> file as the body of the request? I know that's not ideal, but it will test
> my hypothesis.
> > >
> > >
> > > On Thu, Jun 13, 2013 at 9:11 AM, Jeff Minelli <[email protected]> wrote:
> > > Attached are two files, the curl with errors, and the relevant part of
> fedora.log.
> > >
> > > Being desperate, I even tried to blame tomcat myself. :) I've toyed
> with maxPostSize, both using 0 and 4294967296. I even tried the web.xml
> config offered in
> http://tomcat.markmail.org/thread/6jxxcfg2s6rigl6w#query:+page:1+mid:e2l7wd7xhqiwwlpn+state:resultsto
> no success.
> > >
> > > -jeff
> > >
> > > On Jun 13, 2013, at 8:45 AM, Benjamin Armintor <[email protected]>
> wrote:
> > >
> > > > It's possible this is a bug in FCRepo- I'd need to see more of the
> stacktrace, but we've seen issues like this before (
> https://jira.duraspace.org/browse/FCREPO-704). You should also check the
> limits on the maxPostSize in your tomcat configuration, though I would
> expect a different error if that was the problem.
> > > >
> > > >
> > > > On Wed, Jun 12, 2013 at 2:43 PM, Jeff Minelli <[email protected]>
> wrote:
> > > > I have a version of fedora 3.5 running that errors with
> > > >
> > > > javax.ws.rs.WebApplicationException:
> org.fcrepo.server.errors.StreamWriteException: Error writing temp stream
> > > > at
> org.fcrepo.server.rest.BaseRestResource.handleException(BaseRestResource.java:172)
> > > > at
> org.fcrepo.server.rest.DatastreamResource.addOrUpdateDatastream(DatastreamResource.java:583)
> > > > …
> > > >
> > > > when I write a file 2GB in size (1024*1024*2048) via the API:
> > > > curl -H "Content-type:application/octet-stream" -X POST -u
> "name:pass" --upload-file 3g.data \
> > > >
> http://host:8080/fedora/objects/scholarsphere:5425k9692/datastreams/blah?controlGroup=M
> > > >
> > > > But when I write a 2GB-1MB file (1024*1024*2047) via the API, it
> works without issue.
> > > >
> > > > Any ideas?
> > > >
> > > > Thanks,
> > > >
> > > > -jeff
> >
> >
> >
> >
> >
> >
> ------------------------------------------------------------------------------
> > This SF.net email is sponsored by Windows:
> >
> > Build for Windows Store.
> >
> > http://p.sf.net/sfu/windows-dev2dev
> > _______________________________________________
> > Fedora-commons-users mailing list
> > [email protected]
> > https://lists.sourceforge.net/lists/listinfo/fedora-commons-users
> >
> >
> >
> ------------------------------------------------------------------------------
> > This SF.net email is sponsored by Windows:
> >
> > Build for Windows Store.
> >
> >
> http://p.sf.net/sfu/windows-dev2dev_______________________________________________
> > Fedora-commons-users mailing list
> > [email protected]
> > https://lists.sourceforge.net/lists/listinfo/fedora-commons-users
>
>
>
> ------------------------------------------------------------------------------
> This SF.net email is sponsored by Windows:
>
> Build for Windows Store.
>
> http://p.sf.net/sfu/windows-dev2dev
> _______________________________________________
> Fedora-commons-users mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/fedora-commons-users
>
------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:
Build for Windows Store.
http://p.sf.net/sfu/windows-dev2dev
_______________________________________________
Fedora-commons-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/fedora-commons-users
