BibLibre (and Tamil i think) would have the same reports here. From BibLibre experience, main problems are : - utf8 support which in the standard is not supported. You have do disable the error checks. checksum could be badly processed in that case i think.
For what it's worth, I contributed a patch to the original OpenNCIP code which fixed UTF8 checksums with our (at the start of 2010 quite new) 3M self-check machine. The bug report was at http://sourceforge.net/tracker/?func=detail&aid=2925760&group_id=161781&atid=821216 and David Fiander merged the most pertinent bits here: (http://openncip.cvs.sourceforge.net/viewvc/openncip/src/Sip/Checksum.pm?r1=1.4&r2=1.6) A quick peek at Koha's C4::SIP::Sip::Checksum shows that it's still using the %16U instead of the %16C to unpack the packet. Could be an easy fix?
- translations have to be hardcoded for server strings and date strings... Which is quite annoying - Bad dates on renewals when renewal is done (the duedate sticks) and that's it.
I did some work on the Evergreen SIP driver on renewals; the code I wrote probably isn't useful in a Koha context; but it suggests that it can be made to work.
What could be done would be to rework the C4::SIP in order to be more compliant to the License (it was first written with evergreen iirc) and to share more with openncip (openncip.org), where the software originated.
I think that would be great, in spirit if not in domain name (openncip.org doesn't look like it has been updated since 2006, and I think only LibLime has access to that domain). As for the home of the most current code, the CVS repository at SourceForge for openncip still exists, but hasn't been touched for approximately a year. Around May 2010, Joe Atzberger quietly set up a git repository at https://github.com/atz/SIPServer and changed the Evergreen SIP instructions to point to his repository for the base SIPServer code rather than SourceForge; unfortunately his git repository starts with a checkout of the latest CVS code rather than preserving the entire history of the development of the code. Perhaps we could ask David Fiander to convert the SourceForge CVS repository to git within SourceForge to begin with using "git cvsimport" or the like (so that the "official project development home" remains along with the complete repo history, even though it's yucky old SourceForge); then we could begin merging Joe's branch and encourage other contributions. Would that be a reasonable method of revitalizing the OpenNCIP project? -- Dan Scott Laurentian University