On 12 June 2011 13:34, Stefano Rivera <[email protected]> wrote:
> Colin: I see your frustration, but toolkit transitions need to happen. And > nobody (sane) will force a transition, if the packages depending on it > aren't ready or can't be made to be ready in time for release. > thanks for your reply. I don't really follow you. the transition was forced. upstream packages weren't ready. the transition from one library verion to another happened very fast, just before the release of 11.04. there's no way the developers or even packagers could have reacted in time throughout development of 11.04 the old library was in there, from what I can see, then just before release the new one was dropped in. If people had known ahead of time, they could have taken the change into consideration and implemented the (no doubt) minor changes required. in the end I just compiled odin and put it in separate tree. Odin is not the package I'd install. I'd install "mitools". simple command line medical image tools. so it's ok for me, but I don't think it was well handled. many or most scientists are not able to do that easily, or even realise that it is an option. A grad student might help them. > Sure, this will result in some breakage, and some packages may fail to be > ported to the new version, but we can't maintain multiple versions of all > libraries in the archive. > this is a small part of the repository.* most of* the packages that use the library don't work because of the last min library change. of those that do, they are all image viewers. there are a million image viewers for medical data. in fairness, these ones read raw DICOMM files. but not many people really do that any more. the current standard is niftii. so one would convert a dicomm set to niftii for analysis, and read it into FSL, SPM or one of the may other programs in the repositories used as standard by scientists. Brain scientists anyway. except that can't do the conversions. because there are no applications that would convert a dicomm file in the repository. DICOMM was formed in the early eighties. Because of that there are many occasions when it needs to be read or converted. It's also a network protocol. It is a standard. But it's an old standard. older than PERL. the applications are more important than the library. At the moment, the library has no function in the OS that I can see other than to support a few viewers, and it breaks the conversion software to allow the reading of ones data by e.g. FSLView, which used by hundreds of thousands of people. Literally. > In this case, the odin upstreams appear to have been informed about the > issue. > > other packages than odin are affected. > ** Bug watch added: Debian Bug tracker #623145 > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=623145 > > ** Also affects: odin (Debian) via > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=623145 > Importance: Unknown > Status: Unknown > > ** Summary changed: > > - medical imaging is not low priority: about 50% of critical medical > imaging packages are uninstallable by a (possibly unessecarily) changed > dependency libdcmtk1 to libcmtk2 > + Needs porting to DCMTK 3.6 > > ** Changed in: odin (Ubuntu) > Importance: Undecided => Medium > > -- > You received this bug notification because you are subscribed to the bug > report. > https://bugs.launchpad.net/bugs/779170 > > Title: > Needs porting to DCMTK 3.6 > > Status in “odin” package in Ubuntu: > New > Status in “odin” package in Debian: > Unknown > > Bug description: > you have, for some reason, and very recently, removed the package > libdcmtk1 and replaced it with libdcmtk2. the maintainer is mainly > concerned with the ubuntu desktop, not with this sort of package as > far as I can see. > > this means that the packages "odin" and most importantly "mitools" as > well as several other important medical imaging utilities have ceased > to function (i.e. cannot be installed due to broken dependency > libcmtk2 instead of required libdcmtk1). > > libdcmtk is a medical imaging related package. It's basically for > dealing with DICOMM. It doesn't have a wider purpose in the operating > system than managing DICOMM sets. > > I don't understand why it needed to be changed in a way that broke the > dependencies of half the medical imaging software in the repository. > > now, you've listed this as "low priority" > > Even though I'm a just scientist and not a doctor I'd consider things > with the words "medical imaging" in them to be very high priority, > compared to, say, a toolbar or a widget. > > I think you'd agree actually. I need to use miconv to manipulate MRI > images. Since I'm doing basic science, lives are hardly at risk if I > hit a snag. > > However. these same packages are most certainly used by biomedical > scientists and doctors to detect tumors and study and treat disease. > > So I'd put it quite high on the list of things to fix. But it's up to > you. > > I think just putting libdcmtk1 back would do the trick but I've not > traced it through. There must have been some reason to replace it but > too much is now broken. much of my MRI code. > > the medical and neuroscience communities would appreciate these basic > being fixed without us having to mix compiled source and binaries on > the same tree, and not being accorded a "low priority". > > steps required to replaciate : try to install e.g. mitools from > universe repository. > > > Colin > > To manage notifications about this bug go to: > https://bugs.launchpad.net/ubuntu/+source/odin/+bug/779170/+subscriptions > -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/779170 Title: Needs porting to DCMTK 3.6 To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/odin/+bug/779170/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
